SARS-CoV-2-specific T cell responses and immune regulation in infected pregnant women

We studied the T cell response to SARS-CoV-2 spike and non-spike peptide epitopes in eight convalescent pregnant women together with the immune monitoring that included innate tolerogenic dendritic cell populations important to maintain the immunological mother/fetus interface to address a potential risk for the antiviral cellular response in the outcome of pregnancy. Four subjects had pre-existing chronic inflammatory conditions that could have potentially affected the SARS-CoV-2-specific T cell response. Seven of eight subjects responded to SARS-CoV-2 peptides with differences within CD4+ T helper (Th) and CD8+ cytotoxic T cells (CTL). SARS-CoV-2-specific inducible regulatory T cells (iTreg) were numerous in circulation. CD4+ T cell memory included central memory T cells (TCM) and effector memory (TEM). As far as the CD8+ memory repertoire, TCM and TEM were very low or absent in eight of eight subjects and only effector cells that revert to CD45RA+, defined as TEMRA were measurable in circulation. T cells were in the normal range in all subjects regardless of pre-existing inflammatory conditions. The immune phenotype indicated the expansion and activation of tolerogenic myeloid dendritic cells including CD14+ cDC2 and CD4+ ILT-4+ tmDC. In summary, SARS-CoV-2 infection induced a physiological anti-viral T cell response in pregnant women that included SARS-CoV-2-specific iTreg with no negative effects on the tolerogenic innate dendritic cell repertoire relevant to the immune homeostasis of the maternal-fetal interface. All eight subjects studied delivered full-term, healthy infants.


Introduction
Early data suggest that the special population of pregnant women is more susceptible to severe disease when infected with SARS-CoV-2 (Delahoy, 2020;Du Fosse et al., 2020;Ellington, 2020;Zambrano, 2020), and is at increased risk for adverse outcomes such as preterm delivery (Woodworth, 2020). Immune function is uniquely modified during pregnancy to allow for tolerance of the fetal allograft. The immune cells that reside at the interface between the placenta and the uterus are in fact enriched or excluded from the decidua depending upon their regulatory or inflammatory function (Red-Horse et al., 2004;Moffett and Loke, 2006;Mold and Mccune, 2012;Erlebacher, 2013).
Specifically, dendritic cells (DC), the antigen presenting cells (APC) that activate T cells via the presentation of peptides through the human leucocyte antigen (HLA), are very sparse in the decidua. The maternal adaptive immunity that responds to pathogens is "confined" to prevent bi-directional trafficking across the placenta. The immune cells that constitute the maternal-fetal interface are the immune cells that populate the decidua. Specific lineages of suppressive myeloid dendritic cells prevent miscarriage of the pregnancy via IL-10 secretion and by priming regulatory T cells (Treg): tolerogenic dendritic cells (tmDC) defined by the expression of CD4, CD14, the immunoglobulin-like transcript (ILT)-4 with the canonical markers CD11 c and CD11 b (Burns et al., 2013).
Immune response to infection with the novel SARS-CoV-2 virus in pregnancy has not been well characterized.
In a recent case report, it was suggested that an infected pregnant woman had a good antibody response and there was no evidence of fetal damage (Du Fosse et al., 2020). Here we sought to characterize SARS-CoV-2-specific CD4+ and CD8+ T cell responses and the innate and adaptive immune phenotype in a small sample of pregnant women convalescent from SARS-CoV-2 infection.

Study population
The study sample included eight pregnant women ranging in age from 26 to 40 years who were documented to be infected with SARS-CoV-2 at varying times in gestation. Four women were healthy prior to SARS-CoV-2 infection, one was autoimmune diagnosed with polyarthritis rheumatism, one was allergic with atopic dermatitis, and two had ulcerative colitis (Table 1).
Eligible subjects were selected for this study from pregnant women residing in the U.S. who were enrolled in the MotherToBaby prospective cohort study conducted by the OTIS Research Center at UC San Diego (Chambers et al., 2001(Chambers et al., , 2010Chambers et al., 2013;Bandoli and Chambers, 2017;Chambers et al., 2018Chambers et al., , 2019. All enrolled pregnant women provide detailed information about their demographics, health history, and status of SARS-CoV-2 infection through maternal interviews and medical record review. None of the subjects had a severe SARS-CoV-2 infection and hospitalization was not required. The subjects with pre-existing inflammatory conditions were not on steroids medications at the time of the study. The two subjects with ulcerative colitis were on mercaptopurine and anti-α4β7 integrin receptor respectively during pregnancy. Each subject provided a blood sample through a Quest Laboratory closest to their residence, which was subsequently sent by overnight mail to UC San Diego for processing and analysis. The study was approved by the Human Research Protection Program at UC San Diego, and all women provided written informed consent to participate.

Characterization of SARS-CoV-2-specific T cells
All the SARS-CoV-2 peptides were derived by the bioinformatics analyses and synthesized as crude material (T.C. laboratory, San Diego, CA), resuspended in DMSO, and pooled according to megapool (MP) composition followed by sequential lyophilization steps. SARS-CoV-2 CD4 and CD8 peptides MP have been designed and validated in acute and convalescent samples, as previously reported (Grifoni et al., 2020a, 2020b, Mateus et al., 2020Weiskopf et al., 2020). The MP design was carried out on the Wuhan-Hu-1 reference isolated (GenBank Accession Number: MN908947). To ensure a comprehensive assessment of CD4-spike specific reactivity, the main target of vaccine candidates, overlapping 15-mers by 10 spanning the entire protein have been synthetized and pooled separately (Spike; n = 253). The remainder of the SARS-CoV-2 proteome was filtered applying the "7-allele-method" CD4-T cell prediction with a cutoff of ≤ 20, aiming to predict promiscuous epitopes with the capability to bind across the most common HLA class II specificities (Non-spike; n = 221).

Characterization of pro-inflammatory, regulatory and memory T cell responses to SARS-CoV-2
PBMC from the eight subjects were cultured 24 h with peptide pools to study T cell activation, antigen-specific Treg expansion and T cell memory in response to SARS-CoV-2.
The assay involved PBMC separation and 24 -h stimulation with SARS-CoV-2 peptides that included spike and non-spike epitopes for CD4+ T cell responses and spike and non-spike epitopes for CD8+ cytotoxic T cell (CTL) responses. The readout was an activation induced markers (AIM) assay based on flow cytometry to detect the early expression of molecules that define CD4+ and CD8+ T cell stimulation via T cell receptor signaling. Memory T cells on AIM + were enumerated by gating on CD45RA− and further defined as CCR7− effector memory T cells (T EM ) and CCR7+ central memory T cells (T CM ). T EMRA memory cells are effector cells defined by reverting to a naïve CD45RA+

Fig. 1. CD4þ and CD8þ T and CD25 high regulatory T (Treg) cell responses to SARS-
CoV-2 peptide epitopes and the memory phenotypes of the antigen-specific CD4þ and CD8þ T cells. PBMC from eight pregnant women infected by SARS-CoV-2 were separated from heparinized whole blood using Histopaque. PBMC were stimulated with SARS-CoV-2 CD4 spike (253 epitopes), CD4 non-spike (221 epitopes) peptide MPs, or CD8 (MP A and MP B, 314 epitopes each) for 24 h followed by the staining using a combination of monoclonal antibodies to study the T cell and iTreg activation in response to stimulation. The T cell memory phenotype of the antigen-specific AIM+ CD4+ and CD8+ T cells was also studied 24 h after stimulation. T cell and iTreg activation is shown as stimulation index (SI) by calculating the percentage of activated T cells in peptide -stimulated PBMC cultures divide by the percentage of activated T cells in the unstimulated control. A SI ≥ 2 was considered a T cell response to peptide epitopes. Memory phenotype of the AIM + T cells were identified as effector memory (T EM ; CD45RA− CCR7− ), central memory (T CM ; CD45RA− CCR7+) T cells and terminally-differentiated effector that revert to a naïve phenotype (T EMRA ; CD45RA+ CCR7− ). (A) CD4+ and CD8+ T cell responses to SARS-CoV-2 peptide epitopes. All the subjects showed a CD4+ T cell activation in response to SARS-CoV-2 CD4 peptides except for two subjects (#5 and #6). Four subjects (#1, #2, #5, and #6) of the eight subjects also showed CD8+ T cell responses to SARS-CoV-2 CD8 peptides. (B) CD4+ CD25 high iTreg expansion after SARS-CoV-2 CD4 peptides stimulation was found in six (#2, #4 -#8) of eight subjects. (C) Enumeration of SARS-CoV-2specific memory T cells. CD4+ and CD8+ T EM , T CM and T EMRA are shown. Each symbol represents the percentage of T EM and T CM and T EMRA in the AIM+ CD4+ or CD8+ T cell populations. T EM and T CM were measurable in the AIM+ CD4+ T cell population but not CD8+ T cells. Conversely, T EMRA were only AIM+ CD8+ T cells.
phenotype, although antigen experienced T cells. Representative flow images and flow gating strategies can be found in Supplementary Fig. 1. Treg were defined by the expression of CD25 high measured on CD4+ populations stimulated with peptide MPs.
As shown in Fig. 1A, eight of eight subjects showed a T cell response to SARS-CoV-2 with some differences within the Th and CTL repertoire as previously observed in convalescent adults (Grifoni et al., 2020b). Within the four healthy women before infection, only subject #1, had expanded CD4+ T cells specific for spike and non-spike proteins and CD8+ T cells. In this subject, the CD8+ T cell expansion was very pronounced. Coordinated CD4+ and CD8+ T cell responses were also detectable in subject #2: this subject showed only non-spike-specific CD4+ T cells. Subject #4 did not show CD8+ T cell expansion but CD4+ T cells recognized spike and non-spike proteins, as subject #3, that had only CD4+ spike-specific T cells in circulation. Subject #5 with polyarthritis rheumatism and subject #6 with allergy showed only CD8+ T cell responses but not Th responses. In contrast, subjects #7 and #8 with ulcerative colitis had a very pronounced CD4+ T cell response to spike and non-spike proteins but not CD8+ T cell responses.
In pregnancy, regulatory T cells are of the most importance in the maternal-fetal interface. To address the antigen-specific SARS-CoV-2 adaptive regulation, we enumerated CD4+ CD25 high Treg within the AIM+ T cells (Fig. 1B). Eight of eight subjects had numerous spikespecific and non-spike-specific inducible Treg (iTreg) (Fig. 1C and Supplementary Fig. 1), indicating a role in controlling the anti-viral response.
The development of T cell memory plays an important role not only to insure protection for the mother from possible secondary SARS-CoV-2 infection, but also protection of the child during birth and possibly later through maternal immunological microchimera and milk transfer (Hsieh et al., 2019;Franco, 2020). When we examined T EM and T CM by gating in AIM+ T cells, we found differences depending upon timing of infection (Fig. 1C). The main differences were observed between spike-specific CD4+ T EM (numerous in six of eight subjects) and non-spike CD4+ T EM repertoire which occurred at a much lower rate (Fig. 1C). Of note, none of the eight subjects showed a T EM or a T CM CD8+ T cell response irrespective of the timing of infection (Fig. 1C). Within the CD8+ T cell memory populations, T EMRA were inversely numerous within CD8+ T cells but not CD4+ T cells (Fig. 1C). T cell numbers were in the normal range and comparable between subjects with CD4+ T cells ranging from 29.6 to 58.7% (median 46.9 %) and CD8+ T cells ranging from 9.4 to 24.6 % (median 19.8 %). CD4+ and CD8+ CD45RA+ naïve T cells were in the normal range, 54.8-88.9 % (median 79.8 %). Antigen experienced IL-7R+ T cells were comparable between subjects, 42.1-82.6 % (median 67.8 %) (Fig. 2).

The immune phenotype suggested expansion of tolerogenic innate cells
Next, we explored the antigen presenting cell (APC) compartments for T cell presentation including myeloid cells important for the immune tolerance for the protection of the fetus from the same PBMC preparation tested for SARS-CoV-2 T cell responses. The most relevant regulatory innate cells in pregnancy include three myeloid lineages that abundantly secrete the suppressive lymphokine IL-10: CD14+ cDC2 and tmDC defined by the expression of ILT-4 and CD4 that can be found in circulation and tissue resident DC10, defined by the expression of ILT-4 and HLA-G. We included monocytes, macrophages, pro-inflammatory cDC1, CD14− cDC2 and plasmacytoid DC. The expression of CD86 served to define cell activation/maturation in these innate populations. The gating strategies for the analysis of the populations by flow cytometry can be found in Fig. 3A. We also enumerated in circulation tissue resident DC-10, myeloid DC with suppressive function defined by the expression of the HLA-G and ILT-4 in addition to HLA-G+ regulatory decidual macrophages, and CD209+ decidual NK cells. These cells are important for the physiological formation of the villi and the protection of the fetal compartment from pro-inflammatory T cell responses that should not cross the myometrium that could be tracked in the periphery heading to tissues.
Tolerogenic innate cells CD14+ cDC2 and tmDC were numerous in seven of the eight subjects with the exception of subject #8 who had ulcerative colitis, although not statistically significant (Fig. 3B). When we measured the CD86 expression, we found that CD14+ cDC2 and tmDC were fully activated and mature, suggesting that immune tolerance was the most relevant efferent arm of the antigen presentation irrespective of maternal health status before pregnancy and the timing in pregnancy of SARS-CoV-2 infection (Fig. 3B).
Classical APC enumeration within the myeloid compartment included: monocytes, macrophages, cDC1, CD14− cDC2. The results pointed to few differences between subjects and a much lower activation of these cells compared with the tolerogenic counterpart (Fig. 3B). cDC1 and CD14− cDC2, canonical APC for T cell responses, were found in the normal range. Levels of CD86 expression, a marker of activation/ maturity, were high in the two subjects whose samples were collected three weeks after onset of the infection (#1 and #7, respectively), suggesting that the anti-viral T cell response was still ongoing. We also enumerated CD123+ plasmacytoid DC, relevant in the innate response to viruses. These were found to be very low (lower than 6% in total PBMC) and again, not activated in the subjects whose samples were collected close to onset of the SARS-CoV-2 infection (Fig. 3C).
Tolerogenic tissue resident DC-10 and decidual macrophages were measurable in circulation in all the subjects studied with no significant differences in women with inflammatory conditions (Fig. 4A) as CD209+ decidual NK cells (Fig. 4B).

Discussion
Here we report the SARS-CoV-2-specific T cell response and immune phenotype in eight SARS-CoV-2 infected pregnant women. T cells are very important in controlling SARS-CoV-2 infection. It has been an open question about patterns of T cell recognition in infected pregnant women and the integrity of the mechanisms involved in protecting the fetus from pro-inflammatory T cell responses. The fetal HLA differs from maternal HLA and their physiology would suggest an immune response to "allogeneic" antigens. The delicate balance between the necessary pro-inflammatory systemic immune response and strong immune regulation within the placenta occurs through a very complex cross-talk between maternal immune cells and the placental trophoblast. Low numbers and/or functional defects within innate and adaptive cell lineages that regulate the immunity in the maternal-fetal interface contribute to several clinical complications including premature delivery. In physiological conditions, during a viral infection the maternal immune response clears the pathogen, avoiding fetal damage due to pro-inflammatory T cells in the placental interface. A mechanism that prevents T cell activation is the very low number of DC in the decidua that limits the ability to initiate adaptive immune responses in the draining lymph nodes not only to fetal antigens but also to antigens derived from viruses or pathogens (Erlebacher, 2013).
This work reports the T cell anti-viral response in eight SARS-CoV-2 infected pregnant women highlighting the CD4+ and CD8+ recognition of spike and non-spike peptides peptide epitopes. The SARS-CoV-2 proteome was probed using 1102 peptides (9-15 amino acids in length) spanning the whole genome, ensuring the detection of HLA class II-restricted CD4+ T cell responses and HLA class I-restricted CD8+ T cell responses (Grifoni et al., 2020a, 2020b, Lipsitch et al., 2020Mateus et al., 2020;Rydyznski Moderbacher and Dan, 2020;Tarke et al., 2020).
The results indicated that coordinated CD4+ Th cell and CD8+ CTL responses to SARS-CoV-2 were found in subjects closer to the onset of infection (three weeks), and showed greater activation of classical myeloid APC for T cell presentation as cDC1 and CD14− cDC2. This suggested that the T cell response to the virus is still ongoing in subjects  CD11b+ co-expressing ILT-4 and HLA-G) and decidual macrophages (CD14+ CD11c− CD11b+ co-expressing ILT-4 and HLA-G), and their expression of maturation/ activation marker CD86. Uterine DC-10 and decidual macrophages can be found in the PBMC from all the subjects. Subject #5 (polyarthritis rheumatism) and #6 (atopic dermatitis) showed higher percentages of uterine DC-10 then the other subjects with lower percentage of CD86+ cells. Decidual macrophages were found in 0.27 -8.64 % of CD14+ CD11c− CD11b+ macrophage population with variable CD86 expressions among the subjects. Black bars: percentage of different regulatory innate cell populations. Grey bars: percentage of CD86+ in different regulatory innate cell populations. (B) Enumeration of CD209+ decidual NK cells. CD209+ decidual NK cells can be detected in the peripheral blood with a range between 0.01 -0.05 %. No significant difference of the percentage of CD209+ decidual NK cells between the two cohorts was found.
infected months prior to the blood sample collection. Of interest, only the subject with allergy (#6, atopic dermatitis) did not respond to any of the SARS-CoV-2 peptide pools, although the T cell enumeration has not been informative in capturing differences with the other subjects as the number of CD4+ T cells were comparable.
Memory T cells, T EM and T CM , have been found confined to the CD4+ compartment and less numerous in healthy women close to delivery, raising the question of a possible specific homing to the uterus and T cell transfer to the fetus. The absence of T EM and T CM memory CD8+ T cells in circulation but numerous T EMRA (these antigen-experienced effector T cells revert to a naïve CD45RA+ phenotype and are ready to expand upon encounter with the antigen), raises a question on the different timing within the T cell memory development between Th and CTL. This is an important point to better understand the risk of possible reinfection in convalescent subjects.
iTreg that recognized viral epitopes were numerous regardless of the timing of infection, timing of pregnancy or pre-existing conditions. We did not detect Treg in COVID-19 convalescent women (Hsieh et al., 2021) but further investigation is required to drive final conclusions about the unique presence of SARS-CoV-2-specific iTreg in pregnancy. In pregnancy, Treg play a critical role in preventing allogeneic T cell responses and our data indicate that SARS-CoV-2 infection elicits not only a pro-inflammatory T cell response in pregnant women but also a regulatory T cell response. Of interest, two subjects with ulcerative colitis showed the most pronounced response to the virus and the larger number of Treg in circulation. The gut and reproductive system share several lymphoid stations suggesting that women with gut inflammation require further control of pro-inflammatory T cells that may affect the maternal-fetal interface.
The predominant innate populations in circulation were tolerogenic in all the subjects studied: CD14+ cDC2 and tmDC were numerous and highly activated, indirectly suggesting that the immune homeostasis of the maternal-fetal interface is not compromised by SARS-CoV-2 infection. cDC1 and CD14− cDC2, classical APC for T cell presentation, were found in the normal range and activated in women studied close to the time of infection with no significant differences within the subjects.
In summary, SARS-CoV-2 infection does not jeopardize the physiology of the immunity at the maternal-fetal interface. Further evidence is the enumeration of tolerogenic tissue resident innate cells as DC-10 and decidual macrofages that we found in circulation in comparable numbers between healthy women and women with pre-existing conditions before SARS-CoV-2 infection. This is supportive of the current recommendations regarding the safety of vaccination to SARS-CoV-2 any time during pregnancy. Furthermore, in support of this interpretation, all the subjects studied delivered full-term, healthy infants.
Little is known about the placental infection in SARS-CoV-2-infected pregnant women. In two case reports viral load has been found in the placenta associated either with pre-eclampsia at the third trimester (Hosier et al., 2020) or neurological pathology in the baby (Vivanti, 2020) suggesting that healthy deliveries of healthy babies in SARS-CoV-2-infected women does not involve placental infection. More studies are needed to validate this statement: we do not have information in our cohort but the low numbers of pDC, critical innate cells in the anti-viral response and CD14− cDC2 would suggest lack of placental infection.
We acknowledge the strengths and limitations of this work. The sample size of women with inflammatory disorders was small to generalize on all autoimmune and allergic conditions. We first explored the virus-specific T cell repertoire and immune phenotype in SARS-CoV-2 infected pregnant women. However, we did not examine cord blood tissues and blood samples from infants born to the study subjects. This is a limitation in addressing the homing of the T cells, in particular memory T cells, their relation to fetal protection and possible T cell transfer to the baby as a mechanism of protection.

Data availability
No data was used for the research described in the article. Data will be made available on request. All data is within the manuscript and figure The authors do not have permission to share data.