Placental response to maternal SARS-CoV-2 infection

The coronavirus disease 2019 (COVID-19) pandemic affected people at all ages. Whereas pregnant women seemed to have a worse course of disease than age-matched non-pregnant women, the risk of feto-placental infection is low. Using a cohort of 66 COVID-19-positive women in late pregnancy, we correlated clinical parameters with disease severity, placental histopathology, and the expression of viral entry and Interferon-induced transmembrane (IFITM) antiviral transcripts. All newborns were negative for SARS-CoV-2. None of the demographic parameters or placental histopathological characteristics were associated with disease severity. The fetal-maternal transfer ratio for IgG against the N or S viral proteins was commonly less than one, as recently reported. We found that the expression level of placental ACE2, but not TMPRSS2 or Furin, was higher in women with severe COVID-19. Placental expression of IFITM1 and IFITM3, which have been implicated in antiviral response, was higher in participants with severe disease. We also showed that IFITM3 protein expression, which localized to early and late endosomes, was enhanced in severe COVID-19. Our data suggest an association between disease severity and placental SARS-CoV-2 processing and antiviral pathways, implying a role for these proteins in placental response to SARS-CoV-2.


Results
Participant characteristics and maternal-neonatal outcomes. We included in this study 66 COVID-19-positive women: 59 were either asymptomatic or had mild COVID-19, and 7 had severe disease (Fig. 1). As shown in Table 1, women in the asymptomatic/mild disease group were younger than women in the severe group (mean age at delivery for all participants 28.6 ± 5.8 vs 36.0 ± 4.7 years, respectively, p = 0.005). The majority of participants (73%) identified their ethnic origin as Hispanic/Latino/Spanish, with a distribution of other ethnic groups as shown in Table 1. Pre-pregnancy BMI, reported for 44/66 participants, and the incidence of key maternal medical diseases were not significantly different between the groups. The median gestational age at COVID-19 detection and the gestational age at delivery were later for the asymptomatic/mild group when compared to the severe disease group (Tables 1, 2), although the time from symptom onset or disease detection to delivery was insignificantly different ( Table 2).
The maternal outcomes, analyzed by the severity of COVID-19, are presented in Table 2. There were no differences in the mode of delivery. As expected, the use of oxygen and several COVID-19 medications and even the use of IV antibiotics were more common in the severe COVID-19 group. The neonatal outcomes are shown in Table 3 and revealed no differences with respect to neonatal sex, Apgar scores, birth weight, or the rate of NICU admission. The reasons for NICU admission were related to complications of prematurity. All newborns in our study tested negative for SARS-CoV-2.
Placental histopathology. The histopathological features of the placenta, analyzed on the basis of maternal COVID-19 severity, are presented in Table 4. Of the 66 participants, 59 had placental histopathology analysis performed for clinical indications, including all 7 severely symptomatic patients. There were no differences between the two groups with respect to the frequency of maternal vascular malperfusion (MVM) lesions, fetal vascular malperfusion (FVM) lesions, acute inflammatory processes, chronic inflammatory processes, or lesions categorized as not fitting into these groups. Placental histopathologic result of the two cases that were placenta- Placental expression of SARS-CoV-2 entry factors and IFITM. Placental biopsies were obtained from 61 participants. We were able to extract suitable total RNA from 83% of the samples processed in RNAlater and 55% of the samples processed in FFPE (see "Methods"), resulting in suitable RNA samples from 43 participants with asymptomatic/mild (n = 36) or severe (n = 7) COVID-19. Eighteen additional biopsies were www.nature.com/scientificreports/ obtained from placentas of COVID-19-negative women, as described in "Methods". In the absence of adequate specimens for protein analysis, all gene expression measurements were performed using RNA samples, analyzed separately for RNAlater-and FFPE-preserved samples. Using primers specific for the SARS-CoV-2 S-and N-sequences, which we showed to detect ≥ 8 viral RNA molecules (see "Methods"), we detected the presence of SARS-COV-2 in the placenta of two participants, one with asymptomatic/mild and one with severe COVID-19. Using all suitable RNA samples, we measured placental expression of mRNAs for SARS-CoV-2 entry factors,  www.nature.com/scientificreports/ namely ACE2, TMPRSS2, and furin. As shown in Fig. 3, we found that ACE2 levels were lower in asymptomatic/ mild participants compared to those with severe disease. In contrast, we detected no difference in the expression level of TMPRSS2 among participants in the three groups. The expression of furin was lower in participants with asymptomatic/mild disease compared to COVID-19-negative controls. IFITM are innate immune response genes 53 that are expressed in the placenta and are known to limit viral infections 49,54,55 . Located at the cell membrane and endosomes, IFITM proteins can diminish viral fusion with the cell membrane or the fusion of viruses with late endosomal membranes for cytosolic entry 55 . Among the five human IFITM members, only IFITM1, IFITM2, and IFITM3 are known to be interferon responsive 55 . One of the members of this family, IFITM3, is an endosomal protein that was recently shown to restrict SARS-CoV-2 replication 56,57 . We found that placental mRNA expression of IFITM1 and IFITM3 was upregulated in participants with severe disease when compared to asymptomatic/mild COVID-19-positive pregnant women (Fig. 4A). Using immunofluorescence, we found that IFITM1 was localized within the villous core and around fetal capillaries, with weak expression in the trophoblast layer, and no difference was observed between women with asymptomatic/mild or severe disease. The expression of IFITM3 was similar to that of IFITM1, with signal enhancement in specimens from placentas of women with severe as opposed to asymptomatic/mild disease (Fig. 4B). We also showed that IFITM3 colocalized with EEA1 and LAMP1, markers of early and late endosomes, respectively (Fig. 4C). Finally, we found no correlation between the expression of SARS-CoV-2 entry factors and IFITM transcripts and placental histopathology (not shown).

Discussion
Our clinical data corroborated previous observations, indicating that, on the basis of common definitions of infection 58 , fetal transmission is rare, even in women with severe COVID-19. A review of published histopathological analysis of placentas from pregnancies affected by COVID-19 revealed that an association with chronic histiocytic intervillositis was infrequently observed 18,19,24,27,28,59 . Whether or not there is a higher prevalence of maternal or fetal vascular malperfusion lesions remains controversial 28,46,[60][61][62] . Whereas our data support the prevailing conclusion that there are no specific pathological lesions that characterize placentas from COVID-19-infected women, it is clear that reporting bias, lack of controls, and insufficient longitudinal analyses limits our ability to decipher the effect of SARS-CoV-2 infection on placental histomorphology.
Pertaining to the placental expression of SARS-CoV-2 entry proteins, we found increased ACE2 expression in women with severe vs asymptomatic/mild COVID-19. This change was not associated with increased fetal infection. The expression of TMPRSS2 was highly variable, but overall, similar among the groups. The expression of furin was reduced in placentas from women with asymptomatic/mild disease, when compared to negative controls. Our data are consistent with recent observations suggesting that the severity of clinical COVID-19 does not correlate with feto-placental transmission 63 .
The exact post-entry mechanism by which SARS-COV-2 interacts with host cell endocytic pathways remains to be explored. After endosomal processing, the viral genome is released to the cytosol, where it is translated into viral E, M, N, and S proteins, largely in the endoplasmic reticulum, promoting viral RNA replication 64 , with subsequent modifications in the Golgi. Although the remaining steps in SARS-COV-2 processing are not entirely understood, recent data indicate that SARS-COV-2 can usurp lysosomal exit pathways after deacidifying the lysosome's content and deactivating its enzymes 65 . Several interferon-stimulated genes were shown to affect SARS-CoV-2 replication. These include the lymphocyte antigen 6 complex locus E (LY6E), which interferes with SARS-CoV-2 entry into cells by attenuating S-protein-mediated membrane fusion 56,66 and the zinc-finger antiviral protein (ZAP), which limits SARS-CoV-2 replication 67 . In this work, we focused on the evolutionarily conserved IFITM proteins, which are expressed in epithelial cells, including placental trophoblasts, where they restrict the replication of influenza A, SARS-CoV-1, flaviviruses, and several other enveloped viruses 49,50,54,[68][69][70][71] . Unlike other interferon-inducible proteins that interfere with viral replication, IFITM proteins were shown to  www.nature.com/scientificreports/ block membrane fusion of enveloped viruses by using an amphipathic helix domain to alter membrane lipid order and curvature, and inhibit virus-induced fusion pore formation 49 . Notably, in certain cases, IFITM proteins have also been shown to promote cell membrane entry by coronavirus species 48,57,72,73 . Interestingly, the antiviral function of IFITM3 may be attenuated by overexpressed TMPRSS2 56,57 , yet we observed no change in TMPRSS2 expression. In the human placenta, a homozygous SNP rs12252-C mutation in IFITM3 predisposes to SARS-CoV-1 and SARS-CoV-2 infection 71,74,75 . Among IFITM proteins, IFITM3 is also known to have the greatest effect on reducing trophoblast fusion 76,77 . We showed that the expression of placental IFITM1 and IFITM3 transcripts was upregulated in women with severe infection when compared to women with asymptomatic/mild disease or to negative controls. While the expression of IFITM1 and IFITM3 was prominent in the villous core and perivascular regions, IFITM3 was also localized to trophoblasts, and was upregulated in women with severe COVID-19. We also showed that IFITM3 colocalized to early and late endosomes, as shown in other cell systems 70 . Taken together, our data suggest that, in women with severe COVID-19, IFITM1 and IFITM3 proteins participate in trophoblastic immune response and possibly promote placental protection against SARS-CoV-2.
The transport of IgG from the maternal to the fetal circulation is facilitated by the trophoblastic FcRn receptor, usually leading to a higher concentration of IgG in the fetal than in the maternal circulation [78][79][80][81][82][83][84] . Our observation that, in the majority of maternal-fetal dyads, the transfer ratio for anti-NP and for anti-S-trimer IgG is < 1 is consistent with recent data from Atyeo et al. 85 and Edlow et al. 17 , who showed that altered glcosylation status of Fc and bias against the glycoforms of SARS-CoV-2 antibodies reduced IgG transport in maternal-fetal dyads with third trimester COVID-19.
Our work has several limitations. First, our sample size is relatively small, likely reflecting the milder nature of the disease in pregnancy. Second, we recognize that the definitions used for mild or severe COVID-19 are somewhat arbitrary and vary among investigators. We used a common definition that has also been used in reference to pregnant women 12,86-88 . Third, we had no access to well-preserved placental specimens in order to validate our data using western immunoblotting. We have highlighted the inconsistent manner (RNAlater and FFPE) used for preservation of our placental specimens. Indeed, we rejected samples on the basis of RNA quality, likely reflecting extended time between delivery and tissue acquisition. Fourth, all participants in our study were enrolled late in pregnancy, yet the initial infection might have occurred weeks earlier, resulting in a sampling bias. Considering the rapid changes in trophoblast development during pregnancy, conclusions cannot be extrapolated to women infected early in pregnancy. Notably, universal serology testing was not available before July 2020, limiting our ability to confirm the participants' infection status earlier in the pregnancy. Lastly, although our analysis was not contingent upon placental infection status, we recognize that additional measures are needed to validate SARS-CoV-2 infection in the placenta. Nonetheless, our findings add to our knowledge regarding the repertoire of placental cell defense mechanisms [89][90][91] and, specifically, to placental resistance to SARS-CoV-2 infection 27 .

Methods
Participants and specimen procurement. This prospective study included women who presented for labor and delivery at Columbia University Irving Medical Center (CUIMC) between March and October 2020 who tested positive for SARS-CoV-2 by way of PCR assay of nasopharyngeal swab 92 . Women diagnosed with SARS-CoV-2 were classified as asymptomatic/mildly symptomatic women or severely symptomatic on the basis of their symptoms and clinical findings as defined by the National Institutes of Health for non-pregnant adults and adopted by the Society of Maternal Fetal Medicine after modifications that take gestational physiology into account 93 . All participants with COVID-19 received care by the high-risk obstetrical team and neonatal teams at CUIMC. All newborns of SARS-CoV-2-positive mothers were tested for the virus by PCR, using nasopharyngeal swabs in the first 24 h of life. Negative PCR control samples were obtained from both CUIMC and the Steve N. Caritis Magee Obstetric Maternal & Infant (MOMI) Database and Biobank at Magee-Womens Research institute.
The study was approved by the institutional review boards at Columbia University Irving Medical Center (CUIMC, IRB # AAAT0191) and at the University of Pittsburgh (MOMI Databank, STUDY19100240), and all experiments were performed in accordance with relevant guidelines and regulations. All CUIMC's COVID-19 biospecimens were procured and stored at the Columbia University Biobank (CUB), a centralized resource that coordinates the processing, storing, and dissemination of specimens for use in clinical research. Given the time sensitivity of obtaining the samples and the minimal risk to the patient, a deferred consenting model was implemented 94 . All postpartum patients who delivered within the New York Presbyterian Hospital system were contacted after discharge and were offered the chance to participate in the CUB. They provided written informed consent to have their tissues and/or blood stored with the CUB and made available for research. For the purposes of this study, specimens from eligible women were then requested from the CUB. Demographic information was collected, including maternal COVID-19 symptoms and related treatment and obstetric complications. All pregnant women who delivered at UPMC Magee-Womens Hospital in Pittsburgh provided written informed consent to participate in the MOMI Databank for the collection of data and biological specimens for use in biomedical research, under a protocol approved by the University of Pittsburgh.
Blood and placental biopsy specimens were obtained at delivery, whenever feasible, on the basis of clinical conditions and team availability. Maternal and fetal (cord) blood samples were collected from residual clinical samples that were obtained by the Columbia University Biobank. A 0.5 3 cm placental biopsy, performed at a lesion-free midportion of the placenta that was equidistant from cord insertion and periphery 95  www.nature.com/scientificreports/ for formalin fixation and paraffin embedding (FFPE). Placental histopathological analysis was performed, as routinely, by CUIMC's Department of Pathology and Cell Biology, and results were obtained from the health records of consented participants. Placental biopsies from healthy, COVID-19-negative term pregnant women included samples in either RNAlater or FFPE sections. The level of selected   cytokine (IL10, IL17A, IL1β, IL6, IP10, MCP1, MIP1β, TNFα, IL28A, IL28β, and IL29) in the plasma or serum samples were measured using Milliplex human cytokine/chemokine magnetic bead panels (Millipore Sigma, St. Louis, MO) and the Luminex 200 platform (Luminex, Austin, TX). The samples were processed according to the manufacturer's instructions, and cytokine concentrations were quantitated by Luminex xPONENT v3.1 and MILLIPLEX Analyst v5. The intra-and inter-assay precision of these cytokines varied between 1.6-4.4% and 6.7-18.3%, respectively. The mean intra-assay accuracy was 97%. Immunoassays were used to quantify plasma antibodies to SARS-CoV-2 S trimer and nucleocapsid protein (NP) as previously described 96,97 . Briefly, SARS-CoV-2 spike trimer, or NP were coated on 96-well ELISA plate at a concentration of 50 ng/well, respectively at 4 °C overnight. After washing with 0.05% Tween-20 in PBS (PBST), plates were blocked with 300 μl/well of blocking buffer (1% BSA and 10% bovine calf serum in PBS) for 1 h at 37 °C, then washed again with PBST. Antibodies or heat-inactivated plasma samples from COVID-19 patients or healthy donors were serially diluted in a buffer (1% BSA and 20% bovine calf serum in PBS) and then incubated in the plates for 1 h at 37 °C. The plates were then washed with PBST and incubated with Peroxidase AffiniPure goat anti-human IgG (H + L) and goat anti-human IgM antibodies (cat #109-035-003 and 109-035-043, Jackson ImmunoResearch, West Grove, PA, both at 1:10,000 dilution) for 1 h at 37 °C. After a final PBST wash, the antibody binding was detected by incubating with Tetramethylbenzidinsubstrate (Sigma-Aldrich, St. Louis, MO, cat #4444) for 3 min. The reaction was stopped with 1 N sulfuric acid (cat# SA212-1, Thermo Fisher, Waltham, MA). Absorbance was measured at 450 nm and the OD450 values were analyzed using GraphPad Prism 8 (GraphPad, San Diego, CA).

Maternal serum and fetal cord blood antibody and cytokine analysis.
RNA extraction and RT-quantitative PCR. RNA was extracted from placental biopsies in RNAlater using TRI reagent (Molecular Research Center, Cincinnati, OH). The RNA was purified using EconoSpin spin columns (Epoch Life Science, Missouri City, TX). Extracted RNA samples were exposed to RNase-free DNase (Qiagen, Germantown, MD) according to the manufacturer's instructions. RNA was extracted from FFPE samples, using Qiagen's RNeasy FFPE kit (cat #73504) and deparaffinization solution (cat #19093). The quantity and quality of total RNA was determined by a NanoDrop 1000 spectrometer (Thermo Fisher), and selected samples were validated using the Agilent bioanalyzer (Agilent, Santa Clara, CA).
Reverse transcription and quantitative PCR (RT-qPCR) was performed in duplicate, using the ViiA 7 Sequence Detection System (Thermo Fisher) as previously described 98 . For mRNA analysis, total RNA was reverse transcribed using the High-Capacity cDNA Reverse Transcription kit (Thermo Fisher) according to the manufacturer's protocol. Quantitative PCR was performed by means of SYBR Select (Thermo Fisher). For miRNA, cDNA synthesis and qPCR were performed with the miRScript PCR system (Qiagen) according to the manufacturer's protocols. PCR primers are given in Supplementary Table 3. Dissociation curves were run on all reactions. mRNA samples were normalized to the expression of the GAPDH. The fold increase relative to control samples was determined by the 2-ΔΔCt method 99 and compared to SARS-CoV-2-negative controls and was performed separately for specimens that were preserved in RNAlater and those preserved in FFPE.
The presence of SARS-CoV-2 in the placental biopsies was determined using RT-qPCR. As a positive control, we used two lung tissue samples, obtained from two COVID-19 autopsies that were verified using PCR and in situ hybridization at CUIMC and processed in RNAlater. To assess the sensitivity of our PCR assay, we used synthetic RNA transcripts of SARS-CoV-2 N1 and N2 RT-PCR amplicon sequences (Bio-Synthesis, Lewisville, TX), diluted to 100,000 transcript copies/μl in RNA storage solution. This was further diluted to 1000 transcript copies/μl with extracted nucleic acid from human embryonic lung cells, and an extract was used as an internal negative control. We screened two primer sets (Supplementary Table 1) and found that the sensitivity of the N1 set was > 100-fold higher. Using serial dilutions, we determined that our N1 primer set detected the presence of ≥ 8 copies of SARS-CoV2 N1 at PCR cycle < 35. Notably, all positive samples were verified using the SARS-CoV-2 S-protein primers (Supplementary Table 1).
Immunofluorescence staining. For IFITM1 and IFITM3 immunofluorescence, we used paraffin-embedded villous sections from six different placentas (asymptomatic/mild, n = 2, or severe, n = 2, and two negative controls) and stained them with an antibody for IFITM1 (Sigma, cat #HPA004810, 1ug/ml) or IFITM3 (Cell Signaling, Danvers, MA, cat #59212T lot #1, 4.9 ug/ml), detected with the proper secondary antibodies (Donkey Anti-Rabbit Alexa Fluor-594 antibody, Invitrogen, cat #A-21207), and mounted using Vectashield mounting media containing DAPI (Vector Laboratories, Burlingame, CA, cat #H-1200-10). Tissue processed without primary antibodies served as a negative control. Images were captured with an Olympus IX83 inverted microscope and Olympus cellSens software and adjusted using Adobe Photoshop. Expression relative to control was determined by two separate individuals reviewing unmodified images (n > 5 per sample) and scored for presence of strong signal versus weak or absent signal.
Statistics. The analysis first focused on the association of COVID-19 symptom severity (asymptomatic/mild vs severe) with maternal baseline demographic variables, maternal and neonatal clinical outcomes, and histopathology variables. Next, cytokine/antibody associations (classified as high expression vs low/negative cytokine expression, and antibodies as modest vs weak/negative expression) with clinically significant maternal baseline demographic variables and histopathology variables were tested. Observations with missing data were excluded from the relevant analyses. Fisher's exact test was used to analyze categorical and binary variables, and Welch's t-test or exact Wilcoxon-Mann-Whitney test used to analyze continuous variables, as appropriate. Ordinal variables, such as Apgar scores, were also analyzed by exact Wilcoxon-Mann-Whitney test. Normality was assessed via Q-Q plots and histograms. For variables that had a very small sample size for one of the exposure groups (< 10), it was assumed that the distributions of the two groups were similar, and normality was assessed primarily by the larger of the groups. For the cytokine and antibody analyses, we had a relatively small number of participants, mostly from the asymptomatic/mild COVID-19 group. We thus elected a priori to use Fisher's exact test and the exact Wilcoxon-Mann-Whitney test for categorical and continuous variables, respectively, and to only include the participants from the asymptomatic/mild COVID-19 group. Categorical variables for antibody levels (modest, weak, negative) were based upon classifications determined in previous studies using the same assay we previously detailed 96,97 . Categorical variables for cytokine levels were based upon levels above and below the median (high and low, respectively) for each cytokine, based upon available data from the manufacturer. A modified Bonferroni correction was used to adjust for multiple comparisons in the cytokine/antibody analyses. As the levels of different cytokine/antibody variables may be correlated, their tests may also be correlated. To calculate the effective number (M eff ) of independent tests, we used the method proposed by Li et al. 101 to obtain a new significance level of 0.05/M eff . This adjustment to the alpha level was applied to maternal and cord blood cytokine and antibody variables separately. Otherwise, the alpha level was equal to 0.05. All analyses were two-sided and conducted using RStudio version 1.2.5042 (RStudio, Inc., Boston, MA, USA) or SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA). For RNA analysis, fold-change data were analyzed using the Kruskal-Wallis nonparametric test, with post hoc Tukey test for all pairwise comparisons. Analyses were performed using Prism software (GraphPad).