key: cord-0956878-f05kn8kq
authors: Su, Fang-Yi; Zhao, Qingyang Henry; Dahotre, Shreyas N.; Gamboa, Lena; Bawage, Swapnil Subhash; Silva Trenkle, Aaron D.; Zamat, Ali; Phuengkham, Hathaichanok; Ahmed, Rafi; Santangelo, Philip J.; Kwong, Gabriel A.
title: In vivo mRNA delivery to virus-specific T cells by light-induced ligand exchange of MHC class I antigen-presenting nanoparticles
date: 2022-02-23
journal: Science advances
DOI: 10.1126/sciadv.abm7950
sha: 24b6d8990d0432457662eb03c7bfdc792083d81a
doc_id: 956878
cord_uid: f05kn8kq

Simultaneous delivery of mRNA to multiple populations of antigen (Ag)–specific CD8(+) T cells is challenging given the diversity of peptide epitopes and polymorphism of class I major histocompatibility complexes (MHCI). We developed Ag-presenting nanoparticles (APNs) for mRNA delivery using pMHCI molecules that were refolded with photocleavable peptides to allow rapid ligand exchange by UV light and site-specifically conjugated with a lipid tail for postinsertion into preformed mRNA lipid nanoparticles. Across different TCR transgenic mouse models (P14, OT-1, and Pmel), UV-exchanged APNs bound and transfected their cognate Ag-specific CD8(+) T cells equivalent to APNs produced using conventionally refolded pMHCI molecules. In mice infected with PR8 influenza, multiplexed delivery of UV-exchanged APNs against three immunodominant epitopes led to ~50% transfection of a VHH mRNA reporter in cognate Ag-specific CD8(+) T cells. Our data show that UV-mediated peptide exchange can be used to rapidly produce APNs for mRNA delivery to multiple populations of Ag-specific T cells in vivo.

Antigen (Ag)-specific CD8 + T cells express T cell receptors (TCRs) that recognize processed peptide Ags bound to major histocompati bility complex class I (MHCI) molecules expressed on the cell surface. The TCR-peptide MHCI (pMHCI) interaction forms the basis for the exquisite specificity of CD8 + T cell recognition and their cytotoxic activity against target cells bearing cognate pMHCI Ags. This central mechanism has driven increasing interest in delivery approaches that can target and modulate T cells for immu notherapy. Recent studies include delivery of immunomodulatory molecules (e.g., transforming growth factor- inhibitors) using nano particles decorated with antibodies against T cell surface markers, including CD3 and programmed cell death protein1 (PD1), to en hance effector functions within the tumor microenvironment (1) (2) (3) (4) . Programming endogenous CD3 + or CD8 + T cells with polymer/lipid nanoparticles (LNPs) loaded with nucleic acids [e.g., CD45 small interfering RNA (siRNA) and chimeric Ag receptor (CAR)-encoded DNA] has shown potential to silence target genes in T cells or for in situ manufacturing of CAR T cells (5) (6) (7) (8) (9) . The ability to target Agspecific T cells offers opportunities to selectively augment diseaserelevant T cell subsets (e.g., viral or tumor Agspecific T cells) in vivo while maintaining homeostasis and selftolerance of the immune system (10) . To target Agspecific T cells in vivo, strategies include engineered human pMHCI [human leukocyte Ag (HLA)]-Fc fusion dimers to expand human papillomavirus (HPV)-specific CD8 + T cells against HPVassociated malignancies (11) or track virus specific CD8 + T cells by immuno-positron emission tomography imaging (12) , artificial Agpresenting cells composed of pMHCI on nanoparticles or engineered red blood cells to activate Agspecific T cells and enhance their effector function for cancer treatment (13) (14) (15) , tumortargeting antibodies to deliver viral peptides that are cleaved by tumor proteases and then loaded onto MHCI on the tumor cell surface to redirect virusspecific T cells against tumors (16, 17) , and nanoparticles decorated with pMHCII molecules to reprogram autoantigenreactive CD4 + T cells into diseasesuppressing regula tory T cells (T reg ) (18, 19) . These studies highlight the broad appli cations of in vivo delivery to Agspecific T cells.

Despite considerable interest, however, multiplexed delivery to distinct populations of Agspecific T cells remains challenging owing to the complexity of the immune response. For example, more than 500 severe acute respiratory syndrome coronavirus 2 CD8 + T cell epitopes that are restricted across 26 HLA class I alleles have been described so far (20) . Conventionally, pMHCI molecules are expressed by individual refolding reactions to assemble three components-an invariant light chain, a polymorphic heavy chain, and a peptide ligand-into the heterotrimeric structure of endoge nous pMHCI molecules (21) . This serial process precludes produc tion of large pMHCI libraries (22) until the development of peptide exchange strategies mediated by ultraviolet (UV) light (21, 23) , temperature (24) , dipeptides (25) , or chaperone proteins (26) . With UV light-mediated peptide exchange, the heavy and light chains are refolded with a sacrificial peptide containing a photolabile group, such that upon photocleavage by UV light, the sacrificial peptide dissociates to allow an exchange peptide to bind to the MHCI presentation groove (21, 23) . For a particular MHC allele, a single batch of UVsensitive pMHCI molecules can be conventionally refolded and then used to produce hundreds of pMHCI molecules carrying different peptides in one step. For example, pMHCI tetramer libraries with >1000 peptide specificities have been described for the detection of neoAgspecific T cells (27) .

Here, we developed Agpresenting nanoparticles (APNs) syn thesized using UV light-mediated ligand exchange for mRNA delivery to multiple influenzaspecific CD8 + T cells (Fig. 1 ). Our approach increases the precision of T cell delivery compared to antibodies (e.g., CD3 and CD8), is rapidly scalable to different pep tide epitopes, and, through mRNA delivery, will enable a range of applications from in situ manufacturing of T cell therapies to genome editing and regulation (28, 29) . We used UV light-mediated ligand exchange to produce a panel of pMHCI molecules from a sacrificial pMHCI precursor that was sitespecifically modified with a lipid tail. This allowed postinsertion after peptide exchange to preformed LNPs, which was formulated on the basis of a similar DLinMC3DMA based composition as the first U.S. Food and Drug Administrationapproved siRNA drug (ONPATTRO) (30) , encapsulating a model mRNA reporter encoding a camelid single variable domain on a heavy chain (VHH) antibody (31) . We found that APNs decorated with conventionally refolded or peptide exchanged pMHCI mole cules targeted and transfected Agspecific CD8 + T cells in multiple TCR transgenic mouse models in vivo (P14, Pmel1, and OT1) re gardless of the MHC allotype (H2D b for P14 and Pmel and H2K b for OT1). In a mouse model of recombinant influenza A viral in fection (A/Puerto Rico/8/34 H1N1 modified with GP33 Ag, abbre viated as PR8GP33), intravenous administration of a threeplex cocktail of peptideexchanged APNs (NP366/D b , PA224/D b , and GP33/D b ) resulted in the simultaneous transfection of the top three immunodominant PR8GP33-specific T cell populations that was significantly more efficient compared to other major cell popula tions in the spleen and liver. Our data show that UV light-mediated peptide exchange allows for parallel production of APNs for func tional mRNA delivery to multiple Agspecific CD8 + T cell popula tions in vivo.

The insertion of derivatives modified with lipids to preformed nano particles is a wellestablished approach (32) to decorate nanoparticles with ligands that are stabilized by hydrophobic interactions. For example, postinsertion is commonly used to PEGylate liposomes or LNPs using poly(ethylene glycol) (PEG) polymers derivatized with lipid tails (33) . We therefore first sought to express recombinant pMHCI molecules with a sitespecific handle for conjugation of a lipid such that the complex could serve as the starting point for pep tide exchange before postinsertion to LNPs (Fig. 1A) . We expressed and refolded the lymphocytic choriomeningitis virus (LCMV) Ag GP33/D b (KAVYNFATM/D b ) with a Cterminal cysteine in the heavy chain to prevent disruption of native disulfide bonds (34) and retain proper pMHCI orientation for TCR recognition (35) . These were then reacted with 1,2distearoylsnglycero3phosphorylethanolamine (DSPE)PEG2000maleimide to generate lipidmodified GP33/D b molecules. In parallel, we synthesized MC3based LNPs encapsulating enhanced green fluorescent protein (eGFP) mRNA by microfluidic mixers that were characterized by an average diameter of 93.85 nm and a zeta potential of −30.20 ± 0.5 mV by dynamic light scattering (fig. S1, A and B). Postinsertion of lipidmodified GP33/D b pMHCI molecules did not appreciably increase LNP size nor alter the zeta potential (107.9 ± 7.34 nm and −22.73 ± 4.7 mV, respectively) ( fig. S1 , B and C). Successful postinsertion across various APN formulations was confirmed using bicinchoninic acid (BCA) protein quantifica tion (table S1). We also assessed eGFP mRNA concentration using the RiboGreen assay and found that bare LNPs and APNs were com parable (80.20 ± 0.50% to 73.62 ± 0.58%, respectively) (fig. S1C).

We next tested whether GP33/D b APNs can selectively bind to their cognate CD8 + T cells isolated from TCR transgenic P14 mice whose CD8 + T cells express a TCR that specifically recognizes the LCMV GP33/D b Ag ( Fig. 2A) (36) . We found that GP33/D b APNs bound to ~97% of P14 CD8 + T cells, whereas noncognate GP100/D b (KVPRNQDWL/D b ) APNs showed minimal staining (3.22%) (Fig. 2B) . We further tested APN binding using H2K b restricted OVA/K b (SIINFEKL/K b ) APNs and observed similar Agspecific binding when coincubated with their cognate CD8 + T cells isolated from OT1 transgenic mice compared to noncognate NS2/K b (RTFSFQLI/K b ) APNs (Fig. 2C ). The observed 10% crossreactivity in NS2/K b APNs to OT1 CD8 + splenocytes could be due to the binding affinity of H2K b MHC to CD8 coreceptors on CD8 + T cells (37) . We next investi gated whether the binding of APNs to T cells would induce internal ization by T cells given that pMHCI multimers are known to be rapidly taken up by T cells through TCR clustering and receptor mediated endocytosis at physiological temperatures (38) . To do this, we studied the fate of APNs after engaging cognate T cells at 37°C compared to 4°C, which is typically used for pMHCI multimer staining to minimize T cell activation and TCR internalization (39) . We incubated DiIC18(5) solid (1,1'dioctadecyl3,3,3',3'tetramethylindodicarbocyanine, 4chlorobenzenesulfonate salt) (DiD)labeled OVA/K b APNs with OT1 CD8 + T cells at 4° and 37°C, followed by an acidic wash to strip uninternalized APNs bound to the TCRs on T cell surface. We found that DiD fluorescence decreased for cells incubated at 4°C, indicating that APNs remained on the cell surface before acid wash (Fig. 2 , D and E). For cells treated at 37°C, however, we observed no change in fluorescence, suggesting efficient T cell internalization of APNs. By contrast, we found no binding and internalization of noncognate GP100/D b APNs by OT1 CD8 + T cells under all con ditions (Fig. 2E ). To determine whether pMHCIinduced TCR inter nalization could result in functional mRNA delivery to T cells, we incubated P14 splenocytes with GP33/D b APNs loaded with eGFP mRNA. We observed a dosedependent eGFP expression by APN transfected CD8 + T cells [mean fluorescent intensity (MFI), 397 and 506 for 1 and 2g mRNA doses, respectively] in contrast to spleno cytes treated with phosphatebuffered saline (PBS; MFI, 153) or free mRNA (MFI, 118) (Fig. 2F) . Collectively, these results demonstrate that cognate APNs target, bind, and induce T cell uptake for func tional mRNA delivery in vitro.

We next quantified in vivo biodistribution and transfection effi ciency of GP33/D b APNs in TCR transgenic P14 mice (Fig. 3A) . We tested functional delivery to major organs using APNs loaded with firefly luciferase (Fluc) mRNA (Fig. 3 , B and C) and found signifi cantly higher luminescence in the spleens isolated from mice treated with GP33/D b APNs compared to noncognate GP100/D b APNs. Notably, no significant difference was observed in the other major organs between the two groups. To quantify delivery to T cells, we harvested P14 splenocytes 24 hours after infusion of DiDlabeled APNs encapsulating VHH mRNA and observed that cognate GP33/ D b APNs targeted >95% of P14 CD8 + T cells, while GP100/D b APN controls resulted in <2% binding, as quantified by DiD fluorescence (Fig. 3D ). To quantify functional delivery, we used mRNA encoding glycosylphosphatidylinositol (GPI) membrane-anchored VHH as a reporter gene, as it has been shown to achieve durable surface ex pression (>28 days) that can be detected by immunofluorescence staining with antiVHH antibodies (25) . Therefore, we stained for surface expression of VHH and found that GP33/D b APNs resulted in significantly higher transfection efficiency compared to its non cognate counterpart (~40% versus <2%, respectively; Fig. 3 , E and F). Notably, we observed negligible transfection (<2%) of CD8 − spleno cytes in mice treated with either GP33/D b APNs or GP100/D b APNs. Combined with the results that we observed at the organ level by IVIS imaging (Fig. 3 , B and C), our data suggested that the observed Fluc luminescence in the spleen was mainly from the transfection of cognate CD8 + splenocytes. We further confirmed our results in a different model using Pmel mice that have been engineered to ex press the cognate TCR against GP100/D b . Similar to our results in P14 mice, we observed ~30 to 40% in vivo transfection of Pmel CD8 + splenocytes treated with GP100/D b APNs compared to ~3% trans fection with GP33/D b noncognate APNs ( fig. S2 ). Together, these re sults demonstrate that APNs enable T cell targeting and functional mRNA delivery in an Agspecific manner.

to the lack of binding to the noncognate control (GP100/D b APN). (B) Representative flow plots of noncognate APNs and cognate APNs binding to CD8 + T cells in splenocytes from P14 TCR transgenic mice. Frequencies depicted are based on gating on CD8 + cells. (C) Ag-specific binding of cognate and noncognate APNs to CD8 + T cells isolated from TCR transgenic P14 or OT-1 mice. ****P < 0.0001, one-way analysis of variance (ANOVA) and Tukey post-test and correction. All data are means ± SD; n = 3 biologically independent wells. (D and E) OT-1 CD8 + splenocytes stained with OVA/K b APNs at 4° or 37°C and analyzed by flow cytometry before and after treatment with an acetate buffer to strip cell surface proteins. GP100/D b APNs served as a noncognate control. ****P < 0.0001 between OVA/K b APN treatment with and without acid treatment at 4°C; n.s., not significant, where P = 0.61 between OVA/K b APN with and without acid treatment at 37°C; two-way ANOVA and Sidak post-test and correction. All data are means ± SD; n = 3 biologically independent wells. (F) eGFP mRNA expression in P14 CD8 + T cells after coincubation with free-form eGFP mRNA or eGFP mRNA loaded in GP33/D b APNs for 24 hours.

Sacrificial peptides that contain a photolabile amino acid and stabi lize the MHCI complex during refolding have been previously developed for prevalent alleles including H2D b and H2K b in mice (21, 40) . We validated UVmediated peptide exchange by comparing staining of P14 splenocytes using fluorescent GP33/D b tetramers where the pMHCI monomers were either produced by peptide exchange from ASNENJETM/D b (J represents photocleavable amino acid) or conventionally refolded. We first found that tetramers formed with UVlabile peptide present on H2D b MHCI did not cause any nonspecific binding to CD8 + splenocytes isolated from P14 and Pmel mice ( fig. S3 ). After UVmediated peptide exchange into GP33 peptides, UVexchanged tetramers showed comparable staining of CD8 + P14 splenocytes to folded GP33/D b tetramers (>95%), whereas noncognate GP100/D b tetramers produced by refolding or UV exchange resulted in minimal binding ( fig. S4 ). We further tested UVexchanged tetramers to detect endogenous immune responses where T cells have a broad range of Ag specificity and binding affinities to their cognate Ags. To do this, we used the wellcharacterized mouse model of influenza virus PR8 (A/PR/8/34) modified to express the LCMV GP33 Ag (PR8GP33). Infection of mice with PR8GP33 leads to CD8 + T cell responses against at least 16 PR8derived peptide epitopes including NP366 (ASNENMETM/D b ), PA224 (SSLENFRAYV/D b ), PB1703 (SSYRRPVGI/K b ), PB1F2 (LSLRNPILV/D b ), and NP55 (RLIQNSLTI/D b ), as well as against GP33 (41-44), which served as a positive control epitope. We produced a panel of pMHCI tetramers against these six epitopes by peptide exchange (Fig. 4 , A and B) (45) and validated Agspecific splenic T cell responses 10 days after infection (Fig. 4C) (44, 46) .

We next sought to integrate UV exchange for APN production. To do this, we synthesized a panel of three UVexchanged APNs inserted with GP33/D b , GP100/D b , and OVA/K b pMHCI molecules to compare with APNs inserted with pMHCI molecules synthesized using the conventional refolding protocol. In splenocytes isolated from three strains of transgenic mice (P14, Pmel, and OT1), we found that cognate UVexchanged APNs bound to CD8 + T cells similar to the folded APNs (Fig. 4D ). By contrast, we only observed minimal background staining from the noncognate control APNs. Last, we tested whether the UVexchanged APNs can target and transfect virusspecific T cells in vivo using PR8infected mice. To do this, we intravenously injected DiDlabeled PA224/D b APNs to PR8infected mice (44, 46) . At 24 hours after injection, we found that, consistent with the in vitro staining results (Fig. 4D) , both folded and UVexchanged PA224/D b APNs targeted ~80% of PA224specific T cells (Fig. 4E) and resulted in comparable transfection efficiency of the model VHH mRNA (Fig. 4F ).

Antibodies against T cell surface markers, including CD3 and CD8, have been used to target polymeric nanoparticles to T cells in vivo irrespective of Ag specificity (47, 48) . We therefore examined the ability of APNs to transfect virusspecific T cells compared to noncognate cell populations (Fig. 5A) . We focused our analysis on liver and spleen, as these were the major organs that showed APN accumulation after intravenous administration ( fig. S5 ). Flow cytometry analysis of the major cell types [natural killer (NK) cells, B cells, CD4 T cells, dendritic cells, macrophages, monocytes, PA224 + fluspecific CD8 T cells, noncognate PA224 − CD8 T cells, Kupffer cells, hepatocytes, and endothelial cells] revealed that APNs preferen tially transfected flu virus-specific T cells (PA224 + CD8 T cells, 59.46 ± 11.81%) compared to noncognate T cells (PA224 − CD8 T cells, 2.63 ± 2.17%), which comprise a population diversity of approximately 10 6 to 10 8 (Fig. 5B) (49) . As anticipated, transfection was also observed by the reticuloendothelial system, including monocytes and macrophages in the spleen and Kupffer cells in the liver, but at significantly lower transfection efficiency than that of PA224specific CD8 T cells (****P < 0.0001). Compared to cohorts that received PBS, mice that were given folded or UVexchanged PA224/D b APNs resulted in similar transfection levels across all cell populations studied, supporting their equivalency.

To demonstrate simultaneous transfection of distinct Agspecific T cell populations in vivo, we administered a mixture of DiD labeled, conventionally refolded NP366/D b and PA224/D b APNs to PR8infected mice at an mRNA dose of 0.1 and 0.015 mg/kg for each APNs. We found that APNs specifically targeted NP366 and PA224 specific T cells in a dosedependent manner ( fig. S6 ), whereas no detectable DiD fluorescence was observed in NP366 − PA224 − T cells (49) . This resulted in a dosedependent transfection of NP366 + or PA224 + T cells with the model VHH mRNA (Fig. 5C ) compared to minimal transfection (<5%) of NP366 − and PA224 − CD8 + T cells, which was consistent with our earlier observations (Fig. 5B) . Last, to demonstrate multiplexed transfection with UVexchanged APNs, we synthesized and pooled a threeplex panel of APNs presenting the top three immunodominant epitopes (NP366, PA224, and GP33) for PR8GP33 (Fig. 5D ). This APN library efficiently trans fected the three selective clones of T cells with significantly higher expression of the model VHH protein than T cells in mice treated with PBS (Fig. 5E) . Notably, the transfection efficiency across the three T cell clones were comparable with that of PA224/D b APNs administrated alone (Figs. 4F and 5B), suggesting that the APN mediated multiplexed transfection did not compromise the trans fection efficiency in each T cell population tested.

Agspecific CD8 + T cells are key players in adaptive immunity, and their ability to directly kill target cells expressing cognate peptide Ags restricted to MHCI presentation is being harnessed for important applications in cell therapy, vaccines, and autoimmunity (11, 12, (16) (17) (18) (19) . Whereas previous work on delivery to T cells via antibodies against cell surface markers (CD3, CD8, etc.) shows great promise, these markers are expressed by all T cells. Moreover, Agspecific T cell responses are polyclonal (50, 51) ; for instance, across five prevalent HLAA alleles (HLAA*01:01, HLAA*02:01, HLAA*03:01, HLAA*11:01, and HLAA*24:02), more than 110 fluspecific peptide epitopes have been identified for human influenza A virus (PR8) (52) . Therefore, we developed APNs for multiplexed mRNA delivery to Agspecific T cells using UVmediated peptide exchange to expedite produc tion of APNs against a panel of peptide epitopes. Our in vivo data using PR8infected mice showed that APNs with UVexchanged pMHCI molecules transfected PA224specific T cells equivalently to APNs synthesized with conventionally folded pMHCI molecules. This allowed us to construct a threeplex APN library using UV mediated peptide exchange to simultaneously transfect the top three immunodominant T cell populations (NP366, PA224, and GP33 specific) in a mouse model of PR8GP33 flu infection. Notably, while the frequencies of the top three immunodominant T cells in the spleen range from 1 to 4% of the total CD8 + T cells, APNs achieved ~50% transfection efficiency with a model mRNA encoding membraneanchored VHH. This targeting sensitivity has not been demonstrated before with antibody-and chemical compositionmediated targeting. Moreover, the in vivo transfection efficiency of APNs in T cells was comparatively higher than T cell delivery technologies previously reported, including a poly(betaamino ester) based polymeric nanoparticle system functionalized with antiCD3 antibodies (~10 to 20%) (8) and a lipidderived polymeric nano particle system (~1.5%) (53) . However, note that the mRNA used in our study was different from the two studies, which used mRNA encoding CAR and Fluc, respectively.

The use of APNs has the potential to be expanded for more than three peptide epitopes and beyond the two MHC alleles (HD b and HK b ) demonstrated in this study. On the basis of prior studies using the UV exchange technology to generate pMHCI libraries with thousands of peptide epitopes, we expect that UV exchange would be sufficient to produce an APN library with 10 to 20 viral peptide epitopes per MHC allele, the scale that we anticipate in common viral infection settings (e.g., cytomegalovirus, EpsteinBarr virus, and flu) (52, 54, 55) . Moreover, sacrificial UVlabile peptides have been developed for most prevalent HLA alleles, including HLAA*01:01 (STAPGJLEY), HLAA*02:01 (KILGFVFJV), and HLAA*11:01 (RVFAJSFIK) (21, 56, 57) . Therefore, APNs are amenable to other pMHCI molecules, including HLA expressed by human CD8 + T cells. The capability of APNs in transfecting multiple virus specific T cell populations may be used to induce in vivo proliferation n.s., not significant, where P = 0.0606 between PBS and folded APNs (0.03 mg/kg total mRNA dose) in NP366 + flu-specific T cell population; **P = 0.0015 between PBS and folded APNs (0.03 mg/kg) in PA224 + flu-specific T cell population (PA224 + ); ****P < 0.0001; two-way ANOVA and Sidak post-test and correction. All data are means ± SD; n = 3 biologically independent mice. (D) Schematic of multiplexed transfection study. (E) Multiplexed transfection study showing the UV-exchanged APN library transfect three virus-specific T cell populations simultaneously. ****P < 0.0001; two-way ANOVA and Sidak post-test and correction. All data are means ± SD; n = 5 biologically independent mice. of virusspecific T cells to treat virusmediated cancers. For instance, a fusion protein composed of dimerized pMHCI and interleukin2 (IL2) has been developed to expand HPV16 E7 1120 specific CD8 + T cells to treat HPVmediated cancers (11) , and a recent study sug gests that HPVspecific T cells recognizing peptide epitopes derived from HPV E2 and E5 proteins should also be considered to elicit maxi mal tumorreactive CD8 + T cell responses against HPVpositive head and neck cancer (58) .

Other than CD8 + T cells, it may be possible to expand the trans fection capability of APNs to CD4 + T cells by generating APNs with pMHCII. Unlike the cytotoxicity effects triggered by the CD8 + TCRpMHCI interaction, the interaction between CD4 + TCR and MHCII induces the differentiation and proliferation of CD4 + helper T cells and CD4 + T reg cells, which help CD8 + T cell responses and suppress pathogenic autoimmunity, respectively. Therefore, prior studies have developed pMHCIIfunctionalized nanoparticles to expand Agspecific T regs through pMHCTCR interaction for the treatment of autoimmune diseases (18, 19) . However, this is likely more challenging as binding of class II-bound peptides is less stable compared to that of class I-bound peptides (59) , and the interaction between the pMHCII and CD4 + T cells is weaker than that between the class I counterpart and CD8 + T cells (60) . Together, our data support the use of APNs for multiplexed mRNA delivery to virus specific T cells, which can potentially be expanded to transfect broader Agspecific T cell subsets.

Six to 8weekold female mice were used at the outsets of all experiments. P14 [B6;D2Tg(TcrLCMV)327Sdz/JDvsJ], Pmel [B6.CgThy1a/Cy Tg(TcraTcrb)8Rest/J], and OT1 [C57BL/6Tg(TcraTcrb)1100Mjb/J] transgenic mice were bred in house using breeding pairs purchased from the Jackson Laboratory. C57BL/6 for PR8 viral infections were purchased from the Jackson Laboratory. All animal procedures were approved by Georgia Tech Institutional Animal Care and Use Committee (protocol numbers: KwongA100191, KwongA100193, and SantangeloA100169D).

Peptides used for pMHC refolding were synthesized in house using the Liberty Blue Peptide Synthesizer (CEM) and validated using liquid chromatography-mass spectrometry (Agilent). To generate pMHC molecules for bioconjugation, codonoptimized gBlocks for H2D b and H2K b 2m were purchased from Integrated DNA Tech nologies and cloned into pET3a vectors (Novagen). H2D b and H2K b genes were engineered with a Cterminal cysteine by sitedirected mutagenesis (New England Biolabs), and pMHC molecules were expressed and refolded as described previously (23) .

Lipids, including 1,2distearoylsnglycero3phosphocholine), cho lesterol, 1,2dimyristoylracglycero3methoxypolyethylene glyco (DMGPEG), DSPEPEG (18:0 PEG2000 PE), and DSPEPEG2000 maleimide, were purchased from Avanti Polar Lipids. Ionizable lipid DLinMC3DMA was purchased from MedKoo Biosciences Inc. Fluorescent, lipophilic carbocyanine dye DiD was purchased from Thermo Fisher Scientific. LNP was synthesized as described previously (61) . Briefly, lipid mixture containing MC3, DSPC, cholesterol, DMGPEG, DSPEPEG (50:10:38:1.5:0.5 molar ratio), and DiD (1% molar ratio of lipid mix) in ethanol was combined with three volumes of mRNA in acetate buffer [10 mM, pH 4.2, 16:1 (w/w) lipid to mRNA] and injected into microfluidic mixing device Nano Assemblr (Precision NanoSystems) at a total flow rate of 12 ml/min (3:1 flow rate ratio aqueous buffer to ethanol). mRNA encoding eGFP, Fluc, and membraneanchored VHH antibody were gifts from P.J.S. The resultant LNPs were diluted 40× in PBS and concentrated down using Amicon spin filter (10 kDa; Millipore).

To functionalize the synthesized LNPs with pMHC, pMHC was first coupled with DSPEPEGmaleimide and decorated on LNPs via postinsertion (62, 63) . Briefly, a lipid solution of DSPEPEG and DSPEPEG2000maleimide at 4:1 molar ratio was dried under nitrogen and placed in a vacuum chamber for 2 hours to form a thin film. Lipids were rehydrated in PBS at 6.4 mg/ml in a 60°C water bath for 15 min and sonicated in an ultrasonic bath (Branson) for 5 min. Refolded pMHCI monomers with Cterminal cysteine were reduced with TCEP [1:3 pMHC to tris(2carboxyethyl)phosphine hydrochloride (TCEP) molar ratio] at 37°C for 2 hours and mixed with the lipid mixture at room temperature (RT) overnight at 2:1 pMHC/ maleimide molar ratio (64) . Lipidmodified pMHCI molecules were incubated with LNPs at 1:50 maleimide/DLinMC3DMA molar ratio at RT for 6 hours to incorporate pMHCI onto LNPs. The re sultant postinsertion mixture was placed in 1 MDa FloatALyzer (Spectrum) and dialyzed against PBS for 16 hours.

The sizes of APNs in PBS were measured by dynamic light scattering with Malvern Nano ZS Zetasizer (Malvern). Final lipid concentra tion was quantified using a phospholipid assay kit (Sigma Aldrich). The concentration of conjugated pMHCI was determined using a BCA assay kit (SigmaAldrich). The mRNA encapsulation efficiency was quantified by QuantiT RiboGreen RNA assay (Life Technologies) as previously described (7) . Briefly, 50 l of diluted APNs was incu bated with 50 l of 2% Triton X100 (SigmaAldrich) in TE buffer (10 mM trisHCl and 20 mM EDTA) in a 96well fluorescent plate (Costar, Corning) for 10 min at 37°C to permeabilize the particle. Then, 100 l of 1% RiboGreen reagent in TE buffer was added into each well, and the fluorescence (excitation wavelength, 485 nm; emis sion wavelength, 528 nm) was measured using a plate reader (BioTek).

Spleens isolated from P14, Pmel, or OT1 TCR transgenic mice were dissociated in complete RPMI media [RPMI 1640 (Gibco) + 10% fetal bovine serum (FBS; Gibco) + 1% penicillinstreptomycin (Gibco)], and red blood cells were lysed using RBC lysis buffer (BioLegend). CD8 + T cells were isolated using a CD8a + T cell isolation kit (Miltenyi Biotec). For T cell activation, isolated CD8 T cells were cultured in T cell media [complete RPMI media supplemented with 1× nonessential amino acids (Gibco) + 1 × 10 −3 M sodium pyruvate (Gibco) + 0.05 × 10 −3 M 2mercaptoethanol (SigmaAldrich)] supple mented with soluble antimouse CD28 (5 g/ml; BD Pharmingen) and rhIL2 (30 U/ml; Roche) at 1 × 10 6 cells/ml in wells coated with antimouse CD3e (3 g/ml; BD Pharmingen).

P14, Pmel, and OT1 CD8 + T cells (1 × 10 6 cells per sample) were isolated and incubated with APNs (10 g/ml) in fluorescence activated cell sorting (FACS) buffer (1× Dulbecco's PBS + 2% FBS + 1 mM EDTA + 25 mM HEPES) for 30 min at 37°C. Cells were washed three times with 1 ml of FACS buffer before analysis on a 8 of 10 BD Accuri C6. For validation of in vitro transfection, P14 CD8 + T cells were activated for 24 hours as described above and resus pended in T cell media + rhIL2 (30 U/ml; Roche) at 2 × 10 6 cell/ml. Cells (5 × 10 5 ) were coincubated with GP33/D b APN containing eGFP mRNA (1 g) in 24well plates at 37°C. After 4 hours, 700 l of T cell media + rhIL2 (30 U/ml; Roche) was added to each well. After an additional 48hour incubation, cells were washed three times and stained against CD8 monoclonal antibody (mAb; clone 536.7, BioLegend; table S2) at 4°C for 30 min. Cells underwent another two washes with FACS buffer before analysis on BD Accuri C6.

OT1 CD8 + T cells were isolated as described above and incubated with OVA/K b or GP33/D b APNs at 10 g/ml and CD8 mAb (clone 536.7, BioLegend; table S2) at 4° or 37°C for 30 min. Cells were washed with FACS buffer, and a portion of stained cells was analyzed on a BD Accuri C6. The remaining cells were incubated in an acid wash solution (0.5 M NaCl + 0.5 M acetic acid, pH 2.5) for 5 min to strip cell surface proteins as described previously before reanalysis on a BD Accuri C6 (64).

In vivo transfection at organ level using P14 TCR transgenic mice P14 TCR transgenic mice were injected intravenously with GP33/D b or GP100/D b APNs loaded with mRNA encoding Fluc (0.1 mg/kg). Organs were harvested 6 hours after injection and incubated in PBS on ice before IVIS analysis. Organs were soaked in dluciferin solution (2 mM luciferin) in PBS for 5 min. After 5min incubation, bio luminescence images were collected with a Xenogen IVIS Spectrum Imaging System (Xenogen, Alameda, CA). The same type of organs was separated from other organs and imaged together (i.e., spleens from all treatment groups were imaged together).

P14 or Pmel TCR transgenic mice were injected intravenously with GP33/D b or GP100/D b APNs loaded with GPIanchored camelid VHH antibody mRNA (0.2 mg/kg). Splenocytes were harvested 24 hours later and stained against CD8 mAb (clone 536.7, BioLegend), anti camelid VHH antibody (clone 96A3F5, GenScript), and pMHC tetramers (streptavidin, 2 g/ml) on ice for 30 min. The working concentrations of antibodies were listed in table S2. Epitope pMHC tetramers for staining were synthesized in house by mixing bioti nylated pMHC with fluorescently labeled streptavidin at a 4:1 molar ratio (23) . Cells were then fixed with IC fixation buffer (Thermo Fisher Scientific) for the flow analysis (LSRFortessa, BD). All flow data in this study were analyzed with FlowJo v.10 (Tree Star).

PR8 virus was a gift from P.J.S. PR8GP33 was a gift from R.A. (Emory University) and E. J. Wherry (University of Pennsylvania). Six to 8weekold PR8infected C57BL/6 mice were intranasally infected with either PR8 virus or PR8GP33 recombinant virus, as specified in Results and figure captions. PR8infected mice were injected intravenously with NP366/D b and PA224/D b APNs con taining the GPIanchored camelid VHH antibody mRNA (0.03 or 0.2 mg/kg) on day 10 after viral infection. Twentyfour hours after the injection, splenocytes were harvested as described above for immunofluorescent staining. Cells were stained against tetramers (NP366/D b , PA224/D b , 0.2 g of streptavidin/staining sample), CD8a mAb (clone 536.7, BD), NK1.1 mAb (clone PK136, Tonbo), B220 mAb (clone RA36B2, Tonbo), CD4 mAb (clone RM42, Tonbo), and anticamelid VHH antibody (clone 96A3F5, GenScript) on ice for 30 min (65) (66) (67) . Antibodies were all used at 1:100 dilutions, and the specific working concentrations were listed in table S2. Cells were then fixed with IC fixation buffer (Thermo Fisher Scientific) for the flow analysis (LSRFortessa, BD).

Six to 8weekold PR8infected C57BL/6 mice were injected intra venously with conventional or UVexchanged PA224/D b APNs containing the GPIanchored camelid VHH antibody mRNA (0.1 mg/kg) on day 10 after viral infection. Twentyfour hours after the injection, cells from spleen and liver were harvested as described above. Cells were stained against CD8a mAb (clone 536.7, BD), NK1.1 mAb (clone PK136, Tonbo), B220 mAb (clone RA36B2, Tonbo), CD31 mAb (clone PK136, Tonbo), CD45 mAb (clone 30F11, BioLegend), CD4 mAb (clone RM42, Tonbo), CD11b mAb (clone M1/70, BioLegend), CD11c mAb (clone N418, BioLegend), Ly6c mAb (clone HK1.4, BioLegend), F4/80 mAb (clone BM8, BioLegend), and anticamelid VHH antibody (clone 96A3F5, GenScript) on ice for 30 min. Antibodies were all used at 1:100 dilutions, and the specific working concentrations were listed in table S2. Cells were then fixed with IC fixation buffer (Thermo Fisher Scientific) for the flow analysis (LSRFortessa, BD). Cells were identified by a com bination of surface markers: macrophages (CD45 + , CD11b + , CD11c − , and Ly6c − /low), dendritic cells (CD45 + , CD11c + , and CD11b − ), endo thelial cells (CD45 − and CD31 + ), monocytes (CD45 + CD11b + , CD11c − , and Ly6c + ), B cells (CD45 + and B220 + ), CD4 + T cells (CD45 + and CD4 + ), CD8 + T cells (CD45 + , CD8 + , and NK1.1 − ), fluspecific CD8 + T cells (CD45 + , CD8 + , NK1.1 − , and tet + ), NK cells (CD45 + and NK1.1 + ), he patocytes (CD31 − , CD45 − , and F4/80 − ), and Kupffer cells (CD31 − , CD45 + , and F4/80 + ).

Significant differences between control and treatment groups were determined by various statistical analyses. Student's t test was used for twogroup comparison. Oneway analysis of variance (ANOVA) was used for multiplegroup comparison. Twoway ANOVA was used when there were subgroups in each group. Data represent means ± SD in each figure and table as indicated. Statistical analyses were performed using GraphPad Prism 8.0.2 software (GraphPad Software) (*P < 0.05, **P < 0.005, ***P < 0.0005, and ****P < 0.0001).

Supplementary material for this article is available at https://science.org/doi/10.1126/ sciadv.abm7950

View/request a protocol for this paper from Bio-protocol. 

In vivo targeting of adoptively transferred T-cells with antibody-and cytokine-conjugated liposomes

T cell-targeting nanoparticles focus delivery of immunotherapy to improve antitumor immunity

Enhancing adoptive cell therapy of cancer through targeted delivery of small-molecule immunomodulators to internalizing or noninternalizing receptors

Targeting small molecule drugs to T cells with antibody-directed cell-penetrating gold nanoparticles

Systemic gene silencing in primary T lymphocytes using targeted lipid nanoparticles

In situ programming of leukaemia-specific T cells using synthetic DNA nanocarriers

A modular platform for targeted RNAi therapeutics

In vitro-transcribed antigen receptor mRNA nanocarriers for transient expression in circulating T cells in vivo

Interfacing biomaterials with synthetic t cell immunity

Modulation of antigen-specific T-cells as immune therapy for chronic infectious diseases and cancer

CUE-101, a novel E7-pHLA-IL2-Fc fusion protein, enhances tumor antigen-specific T-cell activation for the treatment of HPV16-driven malignancies

In vivo detection of antigen-specific CD8+ T cells by immuno-positron emission tomography

An artificial antigen-presenting cell delivering 11 immune molecules expands tumor antigen-specific CTLs in ex vivo and in vivo murine melanoma models

Engineered red blood cells as an off-the-shelf allogeneic anti-tumor therapeutic

In vivo administration of artificial antigen-presenting cells activates low-avidity T cells for treatment of cancer

Antibody-mediated delivery of viral epitopes to tumors harnesses CMV-specific T cells for cancer therapy

Sensitization of tumors for attack by virus-specific CD8+ T-cells through antibody-mediated delivery of immunogenic T-cell epitopes

Expanding antigen-specific regulatory networks to treat autoimmunity

Peptide-MHC-based nanomedicines for autoimmunity function as T-cell receptor microclustering devices

Comprehensive analysis of T cell immunodominance and immunoprevalence of SARS-CoV-2 epitopes in COVID-19 cases

Design and use of conditional MHC class I ligands

Antigen identification for orphan T cell receptors expressed on tumor-infiltrating lymphocytes

Generation of peptide-MHC class I complexes through UV-mediated ligand exchange

A flexible MHC class I multimer loading system for large-scale detection of antigen-specific T cells

Dipeptides catalyze rapid peptide exchange on MHC class I molecules

High throughput pMHC-I tetramer library production using chaperone-mediated peptide exchange

Large-scale detection of antigen-specific T cells using peptide-MHC-I multimers labeled with DNA barcodes

mRNA-based therapeutics-Developing a new class of drugs

Delivery technologies for T cell gene editing: Applications in cancer immunotherapy

Multi-discipline review: Patisiran. Population PK and/or PD analyses

Engineered mRNA-expressed antibodies prevent respiratory syncytial virus infection

Poly(ethylene glycol)-modification of the phospholipid vesicles by using the spontaneous incorporation of poly(ethylene glycol)-lipid into the vesicles

Delivery of oligonucleotides with lipid nanoparticles

Site-specific antibody-drug conjugates: The nexus of bioorthogonal chemistry, protein engineering, and drug development

Site-selective orientated immobilization of antibodies and conjugates for immunodiagnostics development

In vivo selection of a lymphocytic choriomeningitis virus variant that affects recognition of the GP33-43 epitope by H-2Db but not H-2Kb

Developmentally regulated glycosylation of the CD8alphabeta coreceptor stalk modulates ligand binding

Specificity of CTL interactions with peptide-MHC class I tetrameric complexes is temperature dependent

Phenotypic analysis of antigen-specific T lymphocytes

Discovery of CD8+ T cell epitopes in Chlamydia trachomatis infection through use of caged class I MHC tetramers

Qualitatively different memory CD8+ T cells are generated after lymphocytic choriomeningitis virus and influenza virus infections

The PD-1 pathway regulates development and function of memory CD8 T cells following respiratory viral infection

Microbiota regulates immune defense against respiratory tract influenza A virus infection

Cooperativity between CD8+ T cells, non-neutralizing antibodies, and alveolar macrophages is important for heterosubtypic influenza virus immunity

Immunization with live attenuated influenza viruses that express altered NS1 proteins results in potent and protective memory CD8+ T-cell responses

Quantification of epitope abundance reveals the effect of direct and cross-presentation on influenza CTL responses

Progress and challenges towards targeted delivery of cancer therapeutics

In vivo T cell-targeting nanoparticle drug delivery systems: Considerations for rational design

The many important facets of T-cell repertoire diversity

An unexpectedly large polyclonal repertoire of HPV-specific t cells is poised for action in patients with cervical cancer

Dissection of T-cell antigen specificity in human melanoma

Immune epitope database and analysis resource

Enhanced mRNA delivery into lymphocytes enabled by lipid-varied libraries of charge-altering releasable transporters

Differential evolution and stability of epitope-specific CD8(+) T cell responses in EBV infection

Ex vivo profiling of CD8+-T-cell responses to human cytomegalovirus reveals broad and multispecific reactivities in healthy virus carriers

Conditional MHC class I ligands and peptide exchange technology for the human MHC gene products HLA-A1, -A3, -A11, and -B7

A simple and rapid method for quality control of major histocompatibility complex-peptide monomers by flow cytometry

Functional HPV-specific PD-1+ stem-like CD8 T cells in head and neck cancer

Systematically benchmarking peptide-MHC binding predictors: From synthetic to naturally processed epitopes

Affinity-matured HLA class II dimers for robust staining of antigen-specific CD4+ T cells

Cell specific delivery of modified mRNA expressing therapeutic proteins to leukocytes

A combinatorial approach to producing sterically stabilized (Stealth) immunoliposomal drugs

Conventional versus stealth lipid nanoparticles: Formulation and in vivo fate prediction through FRET monitoring

Synthetic antigen-presenting cells for adoptive T cell therapy

DNA-barcoded pMHC tetramers for detection of single antigen-specific T cells by digital PCR

Individually addressable and dynamic DNA gates for multiplexed cell sorting

Modular nucleic acid assembled p/mhc microarrays for multiplexed sorting of antigen-specific T cells

NIH Cell and Tissue Engineering (CTEng) Training Program 5T32GM8433-30

Accepted 25