T-Cell Therapeutics Targeting Human Parainfluenza Virus 3 Are Broadly Epitope Specific and Are Cross Reactive With Human Parainfluenza Virus 1

Human Parainfluenza Virus-3 (HPIV3) causes severe respiratory illness in immunocompromised patients and lacks approved anti-viral therapies. A phase I study of adoptively transferred virus-specific T-cells (VSTs) targeting HPIV3 following bone marrow transplantation is underway (NCT03180216). We sought to identify immunodominant epitopes within HPIV3 Matrix protein and their cross-reactivity against related viral proteins. VSTs were generated from peripheral blood of healthy donors by ex-vivo expansion after stimulation with a 15-mer peptide library encompassing HPIV3 matrix protein. Epitope mapping was performed using IFN-γ ELIspot with combinatorial peptide pools. Flow cytometry was used to characterize products with intracellular cytokine staining. In 10 VST products tested, we discovered 12 novel immunodominant epitopes. All products recognized an epitope at the C-terminus. On IFN-γ ELISpot, individual peptides eliciting activity demonstrated mean IFN-γ spot forming units per well (SFU)/1x105 cells of 115.5 (range 24.5–247.5). VST products were polyfunctional, releasing IFN-γ and TNF-α in response to identified epitopes, which were primarily HLA Class II restricted. Peptides from Human Parainfluenza Virus-1 corresponding to the HPIV3 epitopes showed cross-reactivity for HPIV1 in 11 of 12 tested epitopes (mean cross reactivity index: 1.19). Characterization of HPIV3 epitopes may enable development of third-party VSTs to treat immune suppressed patients with HPIV infection.

INTRODUCTION Viral infections remain a major cause of morbidity and mortality in patients following hematopoietic stem cell transplantation (HSCT). Respiratory viruses, including human parainfluenza virus, can cause significant pulmonary disease and pneumonias in immunocompromised patients (1). Human parainfluenza virus 3 (HPIV3) is the most common serotype and is detected in 90% of infected patients. For patients with primary immunodeficiency disorders (PIDD) or patients post-transplant, HPIV3 can cause fatal pneumonia in up to 45% of infected patients (2,3). Ribavirin and intravenous gamma globulin have been largely ineffective in lowering mortality, and currently, there are no effective anti-viral therapies nor approved vaccines for parainfluenza infections. Treatment therefore remains primarily symptom-based (4).
Susceptibility to viruses during the post-transplant period is due to compromised T cell immunity prior to full immune reconstitution. Adoptive T cell transfer has been highly successful in restoring viral-specific immunity in this patient population. For the past 20 years, virus-specific T-cells (VST) most commonly targeting cytomegalovirus (CMV) and Epstein Barr virus (EBV) have been used in numerous clinical trials (5,6). Recent advances have led to the development of T cell products that target antigens from multiple viruses. (7)(8)(9) Identification of T-cells recognizing HPIV3 viral proteins led us to develop a Phase I study incorporating HPIV3 antigens in a 6 virus-specific T-cell product (NCT03180216). Pre-clinical data compared HPIV3-specific T-cell responses to four viral antigens (hemagglutinin-neuramidase, fusion, nucleocapsid, and matrix proteins) and identified that matrix, which is also highly conserved across sub-strains of HPIV3, elicited the most robust T-cell response (10). While most studies have used HSCT donors to generate VST, "off-the-shelf" T-cells from partially HLA-matched third party donors have also proven to be effective and reduce the cost and time delay prior to treatment (11,12). Here we sought to identify immunodominant novel epitopes within HPIV3 Matrix protein to optimize the generation of potent HPIV-specific T-cells for adoptive therapy.

Donors
Peripheral blood mononuclear cells (PBMCs) from 10 healthy donors were obtained from Children's National Hospital (Washington, DC) under informed consent approved by the Institutional Review Board in accordance with the Declaration of Helsinki. Eight donors had previously consented blood for virusspecific T cell manufacture for administration to patients enrolled on our IRB and FDA approved Phase I clinical trial evaluating a "first in human" multi-virus specific T cell product (NCT03180216). The remaining two donors were healthy volunteers who consented to our IRB approved blood procurement protocol (Pro00004033).
HLA-restriction of epitope recognition was tested using blocking antibodies against HLA class I and II (Dako Agilent, Santa Clara, CA). To determine the HLA allelic restriction, VSTs were plated at 1 × 10 5 /well with partially HLA-matched phytohemagglutinin (PHA)-treated lymphoblasts (PHA-blasts) either alone or pulsed with peptide (10 mg/ml). Blocking of MHC class I and class II was performed by incubation of VST with polyclonal antibodies to human HLA-A/B/C or HLA-DR/DP/DQ for 1 h prior to plating for ELISpot assay as described above.

Epitope Mapping
VSTs were tested for specificity to individual peptides within the Matrix protein by IFN-g ELISpot. The Matrix 15-mer peptides were pooled using a matrix of 19 combinatorial peptide pools, (Supplemental Figure 1) These pools were designed so that each peptide is only represented in 2 pools. Cross-reactive pools were analyzed, and individual peptides were tested to confirm epitope specificity.

Intracellular Cytokine Staining
A total of 0.5 to 1 × 10 6 VSTs were plated in a 96-well U-bottom plate and stimulated with pooled pepmixes or individual peptides (200 ng/peptide/well) or actin (irrelevant peptide control) in the presence of Brefeldin A (Golgiplug; BD Biosciences, San Jose, CA) and CD28/CD49d (BD Biosciences) for 6 h. T-cells were fixed, stained with fluorophore-conjugated antibodies against CD4, CD8 and TCRab, permeabilized with Cytofix/Cytoperm solution (BD Biosciences), and stained with IFN−g and TNF-a antibodies (Miltenyi Biotec).

Data Analysis/Statistics
Results were evaluated using descriptive statistics (means, medians, standard deviations, ranges). Analysis was performed in GraphPad Prism (GraphPad software, La Jolla, CA).

HPIV3 Epitopes Demonstrate Cross-Reactivity With HPIV1
Given the homology between HPIV3 and related Respirovirus Human parainfluenza virus 1 (3), we sought to determine if immunodominant epitopes identified within HPIV3 Matrix protein were cross reactive with corresponding Human Parainfluenza Virus 1 epitopes. We initially tested overall specificity of HPIV3 specific T cell products for HPIV1 Matrix pepmix compared to HPIV3 Matrix pepmix using anti-IFN-g ELISpot (Supplemental Figure 3). The mean response from all 10 products elicited by HPIV3 pepmix was 319.8 SFU/1x10 5 cells, while the mean response elicited by HPIV1 pepmix was 138.9 SFU/ 1 x10 5 cells. Median actin, used as negative control, was 0.5 SFC/ 1×10 5 cells (range 0-1). This resulted in a cross-reactivity index (HPIV1 response/HPIV3 response) of 0.4. We evaluated the crossreactivity of HPIV3 specific T-cells against HPIV1 peptides using anti-IFN-g ELISpot. Cross-reactivity indices (HPIV1 mean SFU/ HPIV3 mean SFU) ranged from 0.03 to 4.5 (Table 3). There was a high level of cross-reactivity between HPIV3 Peptides 83-84 and HPIV1 Peptides 82-83, with a cross-reactivity index of 1 and 1.07, respectively. HPIV1 Peptides 37 and 38 elicited a stronger response from HPIV3 specific T-cells than the corresponding HPIV3 epitopes, with cross-reactivity indices of 2.71 and 4.5, respectively. Of the 12 identified epitopes, 7 to 13 out of 15 amino acid sequences were conserved between strains ( Figure 3A).

HPIV3 Epitopes Elicit an MHC Class II-Restricted Response
HLA blocking antibodies were used in the presence of HPIV3 specific T-cells and peptides 82, 83, and 84. IFN-g production by ELISpot was decreased by a mean of 56.5% (range 36.3%-80.4%) with anti-HLA-DR antibodies. Restriction analysis of peptides 82, 83, and 84 using anti-IFN-g ELISpot showed that VSTs responded to peptide-pulsed PHA blast targets matched at HLA-DRB1*15:01, with mean SFU/1x10 5 cells of 137.7, 97.5, and 99.7 (when presenting 82, 83, and 84, respectively) compared to PHA blasts and peptides alone (mean SFU/1x10 5 cells of <1) ( Figure 4A). This was in line with in silico predictions of MHC binding of the peptides. In comparison, restriction    analysis was performed using VSTs and PHA blast targets partially matched at HLA-DRQ1*03:01. VSTs did not respond to peptides presented by these PHA blasts, with mean SFU/1x10 5 cells decreased by 93%-95%.

DISCUSSION
Human parainfluenza virus infections in immunocompromised patients remain a significant cause of morbidity and mortality and lack specific anti-viral treatment. Virus-specific T-cells have proven effective in treatment of several viral infections occurring post-stem cell transplantation and our preclinical studies have shown that HPIV3 specific T-cells can be produced from healthy donors and successfully target HPIV3 epitopes (10). Further defining VSTs targeting HPIV3 will expand their applicability as third party products. Because transplant donor-derived VST production is limited when the donor is unrelated or the stem cell source is cord blood, and it takes many days to generate the product, "off-the-shelf" VST could be used as immediately available products to treat a number of viruses (11,15,16). However, this approach has not yet been applied to HPIV3 infections. Hence, identification of immunodominant HPIV3 epitopes is essential for selection of partially-HLA matched VSTs, and development of third-party HPIV3 VST therapy.
In this study, we identified 12 new epitopes within the HPIV3 Matrix protein, which were recognized by multiple donors. We discovered a highly dominant epitope within the group, GEFRYYPNIIAKGVG, which elicited anti-viral activity from all 10 products. Overlapping 15mers AIFQPSLPGEFRYYP and PSLPGEFRYYPNIIA also produced responses to 7 and 9 out of 10 products, respectively. This cluster of epitopes resides at the C-terminus, containing a leucine residue critical for regulating virion release ( Figure 4B) (17). Although there has been limited success to date with parainfluenza vaccinations, the identification of this immunogenic epitope cluster may also have significance for future vaccine development (18,19).
We determined that the identified epitopes are HLA class II restricted, similar to the predominance of CD4-restricted responses against the adenovirus Hexon protein and BK virus large T antigen (11,12,20,21). These findings reiterate the importance of CD4 + T cells in their role of respiratory virus clearance. Epitopes induced polyfunctional responses based on TNF-a and IFN-g production. In silico predictions suggested that the most immunodominant peptide, GEFRYYPNIIAKGVG, would elicit a T cell response restricted through HLA-DRB1*15:01. Only 2 out of 10 products were matched at that allele, but restriction analysis showed that the VSTs responded to PHA blast targets matched at HLA-DRB1*15:01. We also showed that IFN-g production was decreased by 32%-80% in the presence of HLA-DR blocking antibodies. Similar findings were noted in CD4+ responses against MHC Class II-restricted adenovirus Hexon epitopes, where the T-cell response of multiple T cell products with differing HLA-alleles was blocked by an HLA-DR antibody (22). Our findings suggest that while there is some degree of HLA-DR restriction, T cell activity against GEFRYYPNIIAKGVG is likely not limited to a single Class II allele. Although our products were manufactured from a diverse group of donors, a limitation to this study is that we cannot evaluate for all possible HLA restricted responses. However, we and others have shown that even knowing a relatively restricted number of class I and class II restricted virus-specific responses provides important clinical information when evaluating virus-specific T cell products in a third party "off the shelf" setting (16,23).
Previous studies have shown that cytotoxic T-cells stimulated by HPIV1-infected cells also recognize HPIV3 proteins (24). Additional studies showed that while antibody cross-reactivity is minimal within the paramyxovirus family, T cell cross-reactivity is common between HPIV3 and other related viruses such as measles, mumps, and RSV (25). Since HPIV3 and HPIV1 both occur in an immunocompromised population, we tested peptide sequences from HPIV1 corresponding to the amino acid sequences from the identified HPIV3 epitopes, and confirmed cross-reactivity in 10 of 12 epitopes. Two of the HPIV1 peptides elicited a more robust T cell response than their homologous HPIV3 epitopes in one subject, which suggests that the elicited T-cell response from the donor may have been derived from a recent immunologic response to HPIV1 rather than to HPIV3. The highest degree of cross-reactivity was between HPIV3 epitopes 83 and 84 and the corresponding HPIV1 epitopes (82 and 83, respectively), which were the epitopes recognized most consistently by the T cell products. Though the clinical impact of T-cells targeting these identified epitopes is not yet known, their immunodominance and cross-reactivity with HPIV1 argue that T-cell response to these epitopes serves an important role in the adaptive immune response to human parainfluenza viruses.
In conclusion, we have characterized multiple novel immunodominant epitopes within HPIV3 Matrix protein. These HPIV3 epitopes show a high degree of cross-reactivity with corresponding HPIV1 epitopes. Our study paves the way for developing third party HPIV3 and HPIV1-VST products to rapidly treat immunocompromised individuals infected with these lethal respiratory viruses.

DATA AVAILABILITY STATEMENT
All datasets presented in this study are included in the article/ Supplementary Material.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by: the Children's National Hospital IRB (Protocol IRB # Pro00008637). Written informed consent to participate in this study was provided by the participant, or participants' legal guardian.

AUTHOR CONTRIBUTIONS
KH, SH, VK, MJ-W, GS, HL, CL, and AD contributed by performing lab experiments including cell culture, intracellular cytokine staining flow cytometry and ELISpot, as well as assisted with data analysis. MK and CB contributed by providing mentorship, laboratory leadership, experiment design, and data analysis. All authors contributed to the article and approved the submitted version.

ACKNOWLEDGMENTS
We thank our donors for participation in this study, as well as the staff of the Center for Cancer and Immunology Research at Children's National Hospital.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu.2020. 575977/full#supplementary-material SUPPLEMENTARY FIGURE 1 | Combinatorial peptide pools HPIV3 Matrix peptides were pooled according to this matrix. Cross-reactive pools were analyzed and individual peptides were tested to confirm epitope specificity, as each peptide is uniquely represented in two specific pools.
SUPPLEMENTARY FIGURE 2 | Gating Strategy. Compensation was performed using antibody capture beads (all antibodies) and cells (viability dye). Cells were acquired on a Beckman Coulter CytoFlex S using CytExpert version 2.2.0.97 software. Data was analyzed on Flow Jo version 10.5. Cells are first gated as singlets, then live cells. CD8+/CD4+/CD3-/CD3+CD4-CD8-cell were interrogated identically using bivariate plots assessing IFN-g and TNF-a. Where > 8% of events fell on an axis bi-exponential scaling was used to visualize all cells on the plot.
SUPPLEMENTARY FIGURE 3 | Overall HPIV3 and HPIV1 Matrix specificity by product All 10 viral-specific T cell products were stimulated with both HPIV3 pepmix and HPIV1 pepmix. Response was measured as spots per well (SFU/1x10 5 cells) by anti-IFN-g ELISpot assay. Unstimulated T-cells (CTL only) and stimulation with actin pepmix (irrelevant peptide) were used as negative controls.