LCMV-Armstrong infection of mice elicits a robust T cell response that drives clearance of the virus by 8-10 days after infection, leaving behind a large pool of quiescent memory CD8⁺ T cells that persist for the life of the animal (28). To determine the impact of a pre-existing, quiescent memory T cell population on the early allogenic response, we developed a murine heart transplant model in which we can track the direct activation of these memory T cells against the allograft. Heart allografts from Balb/c mice were heterotopically transplanted into Nur77-GFP transgenic C57BL/6 recipient mice that were either LCMV- naïve or LCMV-immune, which were given LCMV 8-10 weeks prior ( Figure 1A ). On days 1 through 4 after transplant, hearts were collected and CD8⁺ T cell infiltration was assessed by flow cytometry ( Figure 1A ). As early as day 1 after transplant, LCMV-immune mice had a remarkable infiltration of CD8⁺ T cells into the allograft ( Figure 1B , Supplemental Figure 1 ). Overall, there were more infiltrating CD8⁺ T cells in immune compared to naïve mice on days 1-4 post-transplantation, with significant differences seen at days 1 and 2 after transplantation ( Figure 1B ). These data show that immune mice have increased kinetics and magnitude of CD8⁺ T cell infiltration into the allograft relative to naïve mice.

To confirm the flow cytometric data, we performed histological analyses via H&E staining on transplanted Balb/c hearts. Histologic staining showed progressively increased lymphocytic infiltration from day 1 through 4 after transplantation, which was substantially more pronounced in immune mice ( Figure 1C ). Blinded histologic analyses and scoring on H&E-stained tissue sections reflected the flow cytometric data as well, with earlier and progressively increased histologic rejection scores seen in the immune mice ( Table 1 ). For example, rejection scores were more severe in immune mice at day 2 as compared to naïve mice at day 3. Further, histologic analyses revealed that immune mice had infiltrating neutrophils and eosinophils as early as day 2 after transplant, indicating worsened pathology as compared to naïve mice at day 4 post-transplant, which only had a few neutrophils and eosinophils present. Naïve mice had small foci of epicardial damage at days 1 and 2 after transplant which is expected due to the ischemia/reperfusion injury inherent in the heart transplant procedure. While the overall histology scores were the same by day 4 after transplant, the immune mice had more severe myocyte tissue damage and more inflammatory infiltration ( Table 1 ). Overall, the presence of large numbers of memory T cells prior to transplantation accelerates both the recruitment of CD8⁺ T cell into heart allografts and allograft tissue damage.

Taking advantage of the Nur77-GFP transgenic mice, we next assessed GFP expression in the graft-infiltrating CD8⁺ T cells. Given that GFP expression is limited to those CD8⁺ T cells receiving very recent TCR signals, we considered GFP⁺ T cells in the allograft to be alloreactive. On average, approximately 10% of the CD8⁺ T cells in the graft were alloreactive (GFP⁺) from days 1 through 4 after transplant in both naïve and immune mice ( Figure 2A ). Although no significant difference was seen in the percentage of alloreactive CD8⁺ T cells recruited to the allograft between the naïve and immune mice, immune mice had dramatically increased numbers of alloreactive cells due to a more robust CD8⁺ T cell response ( Figure 2A ). Expression of CD69, measured as an independent marker of activation, was also found to be increased in the number, but not the frequency, of CD8⁺ T cells in immune mice ( Figure 2B , Supplemental Figure 2 ). Given the transient nature of Nur77-GFP and CD69 expression, the extent of early activated CD8⁺ T cells detected is likely an underestimate.

To determine whether this increased infiltration was explained by an increased presence of memory T cells, we examined expression of CD44 on graft infiltrating CD8⁺ T cells. Although expression of CD44, in this instance could also reflect activation within the allograft, we found that significantly less CD8⁺ T cells in naïve mice were CD44^hi (40-80%), while the majority of the CD8⁺ T cells in immune mice were CD44^hi (80-90%) ( Figure 2C ). Consistent with their delayed response, the frequency and number of CD44^hi cells in naïve mice increased from day 1 to day 4 ( Figure 2C ). Additionally, we saw that, compared to days 1 and 2, at days 3 and 4 after transplantation, there was a large increase in the percentage of CD8⁺CD44^hi cells after transplantation in naïve mice, indicating the time it takes for cells to transition from CD44^lo to CD44^hi after antigen stimulation ( Figure 2C ). Further, recruitment of CD44^lo cells into allografts of naïve mice persisted for days, and the overall numbers of CD44^lo and CD44^hi cells were both increased in immune mice ( Figures 2C, D ). Overall, there was preferential early recruitment of memory CD8⁺ T cells into the allograft in mice with a prior LCMV infection.

Next, we determined whether allograft infiltrating CD8⁺ T cells are recognizing alloantigen in the graft, by assessing expression of GFP as an indicator of TCR stimulation in those CD8⁺ T cells in relation to CD44 expression. Interestingly, we found minimal differences in the percentage of CD8⁺CD44^hi that were GFP⁺ between naïve and immune mice, although the overall numbers of CD8⁺CD44^hiGFP⁺ were significantly increased early after transplant in immune mice ( Figure 2E ). Further, immune mice also had an increase in CD8⁺CD44^lo cells that were GFP⁺ compared to naïve mice ( Figure 2F ). As the level of GFP per cell is proportional to TCR signal strength, we assessed the mean fluorescence intensity of GFP in CD8⁺ T cells in naïve vs immune mice (27, 29). We focused on comparing CD44^lo T cells in naïve mice (which should represent the naïve T cell response to alloantigen) to CD44^hi T cells in immune mice (which, at least in the first two days, should be largely comprised of cross-reactive T cells). In general, we found that the CD44^lo T cells in naïve mice had a higher GFP MFI compared to CD44^hi T cells from immune mice, suggesting that naïve T cells responding to alloantigen are receiving stronger TCR signals ( Figure 2G ). Thus, in addition to recruitment and activation of pre-existing memory CD8⁺ T cells into heart allografts, immune mice also had increased numbers of allo-activated naïve CD8⁺ T cells, although they may be of lower affinity.

To assess whether the increase in graft-infiltrating CD8⁺ T cells in LCMV-immune mice is an alloresponse or a non-specific overall response due to inflammation associated with ischemia-reperfusion injury in organ transplantation, immune mice were subjected to either a syngeneic (C56BL/6) or an allogeneic (Balb/c) heterotopic heart transplantation, and samples were collected on days 1 and 2 post-transplantation ( Figure 3A ). We observed more CD8⁺ T cell infiltration in mice that received allogeneic transplants relative to those receiving syngeneic transplants even as early as days 1 and 2 after transplant ( Figure 3B ). Further, we found a significant increase in the frequency of CD8⁺ T cells expressing GFP in mice receiving allogeneic, relative to syngeneic, heart transplants ( Figure 3B ). Further, overnight culture with IL-2 failed to drive an increase in GFP expression in sorted memory CD8⁺ T cells, whereas culture with Balb/c T cell-depleted splenocytes induced a significant increase in GFP expression ( Figure 3C ). Furthermore, allostimulation in vitro also induced degranulation, via CD107a expression, in the GFP population of sorted memory CD8⁺ T cells ( Figure 3D ). Again, stimulation with IL-2 did not result in CD107a expression ( Figure 3D ). Together, these data demonstrate that increased expression of GFP within LCMV-specific CD8⁺ T cells is likely due to TCR activation by alloantigens. Importantly, similar analyses performed in the lymph nodes of these mice reveal no significant differences in CD8⁺ T cell activation ( Supplemental Figure 3 ), confirming the allograft-specific increase in CD8⁺ T cell GFP expression.

To verify that the recruited memory cells included those with LCMV specificity, we stained graft-infiltrating CD8⁺ T cells with two LCMV-specific tetramers, D^bGP33 and D^bGP276. No significant differences were seen between the proportion D^bGP33⁺ and D^bGP276⁺ CD8⁺ T cells ( Figures 3E, F ), despite a trend towards higher frequency and numbers of D^bGP33⁺ CD8⁺ T cells seen at day 1 after allotransplantation ( Figure 3E ). Together, these data show that prior viral infection elicits viral-specific memory T cells that have the capacity to recognize alloantigens in the heart allograft and that their initial recruitment is preferential for cells with individual LCMV specificities.

Next, we determined whether graft-infiltrating LCMV-specific memory CD8⁺ T cells were capable of recognizing alloantigens on transplanted hearts by assessing their expression of Nur77-GFP. As expected, nearly all the LCMV-tetramer positive cells were CD44^hi, indicating their memory status, and already by day 1 post-transplantation, there was a significant increase in CD44^hiGP33⁺ cells seen in allogeneic hearts ( Figure 4A ). The fidelity of the Nur77-reporter was confirmed as a substantial fraction in allogeneic hearts were GFP⁺, while virtually none expressed GFP in syngeneic hearts ( Figure 4A ). Thus, in addition to their preferential recruitment to allogeneic heart grafts, a portion of these cells were also re-activated in the allograft, demonstrating their cross-reactivity to alloantigen in vivo.

Intriguingly, when we assessed recruitment and activation of D^bGP276-specific CD8⁺ T cells, we observed far fewer GFP⁺ cells in allogeneic hearts ( Figure 4B ). Although there were no significant differences seen in the percentage of D^bGP33- and D^bGP276-positive cells in the CD8⁺CD44^hi populations ( Figure 4C ), D^bGP33⁺ cells were proportionally more activated based on GFP expression compared to D^bGP276⁺ cells, with significantly more of them infiltrating into heart allografts 2 days after transplantation ( Figure 4D ). Furthermore, when we assessed the mean fluorescence intensity (MFI) of the GFP⁺ populations of D^bGP33-and D^bGP276-positive cells, we observed that there was a significantly higher MFI in the D^bGP33⁺ cells at day 2 after transplantation ( Figure 4E ), indicating increased TCR signaling strength in the D^bGP33 cross-reactive cells. Combined, our data show that prior viral infection elicits significant increases in allograft infiltration by viral-specific T cells, and these viral-specific T cells display epitope-specific variability with their ability to cross-react with alloantigens.

To ascertain the level of similarity among the TCRs from the different clonotype populations and identify any features in the CDR3 sequences distinguishing the potentially alloreactive clonotype populations from those that are not alloreactive, we sorted, sequenced, and analyzed the properties of CDR3 sequences from the following clonotype populations: CD44^lowGFP^-, CD44^lowGFP⁺, GP33⁺GFP^-, GP33⁺GFP⁺, GP276⁺GFP^-, and GP276⁺GFP⁺.

We first examined the length and compositions of CDR3 sequences, including fractions of acidic/basic residues, hydrophobic residues, and aromatic residues ( Supplemental Figure 4 ). We found that, for both the CDR3α and CDR3β sequences, although the degree of variability in each population differed, the median length was similar across all populations, differing by not more than 1 amino acid (excluding GP276⁺GFP⁺, for which only one paired sequence was obtained). Similarly, no clear differences were observed in amino acid composition between the GFP^- and GFP⁺ populations. We did notice that the median value for the percentage of acidic residues in the CDR3β sequences of the GP33⁺ populations (both GFP⁺ and GFP^-) was greater than other populations, possibly reflecting differing antigen specificity rather than alloreactivity. A similar trend was observed for the percentage of aromatic residues.

For a more granular analysis, we constructed sequence logos using the CDR3 sequences of TCRs from the different clonotype populations (excluding GP276⁺GFP⁺ population as it contained only one paired CDR3 sequence). Here, notable differences in the CDR3 sequences were observed. For example, we observed that the CDR3α sequences of TCRs from the CD44^loGFP⁺ clonotypes, unlike those from the CD44^loGFP^- clonotypes, were enriched in a distinctive GYSGG motif in their core ( Figure 5A ). In contrast, the CDR3β sequences of most of the TCRs from both CD44^lo populations share a glycine in the middle of their sequences ( Figure 5B ). We also noticed differences between the non-viral reactive and viral reactive populations. Although the CDR3α sequences of most of the TCRs from viral reactive (GP33⁺ and GP276⁺) clonotypes and most of the TCRs from the CD44^lo clonotypes share the CA(A/L) and KLTF motifs at the beginning and end of their sequences, they differ in the amino acids found in the middle of the sequences, with G, N, Y and G, Y, G, N being prominent in GP33⁺ and GP276⁺ populations, respectively ( Figure 5A ). The conserved CA(V/A)S motif in the CDR3α sequences of the TCRs from the GP33⁺GFP^- and GP33⁺GFP⁺ clonotype populations suggests shared Vα-gene usage. However, less conservation is observed at the region of the sequences formed by the Jα segment. There is also an enrichment of a G/SNN motif in the CDR3α sequences from the GP33⁺GFP⁺ population ( Figure 5A ). Both populations also differ in their CDR3β sequences, with most TCRs from GP33⁺GFP⁺ population having D, W, D, and G in the middle of their CDR3 sequences ( Figure 5B ). Thus, it appears that cross-reactive CD8⁺ T cells are enriched in populations of cells whose TCRs have distinctive sequences that may endow their cross-reactivity.

7-8 week old Nur77-GFP transgenic mice (C57BL/6 background) were infected with 2x10⁵ pfu of LCMV via intraperitoneal injection. Uninfected Nur77-GFP mice were used as control. Approximately 8 weeks later, mice received either a syngeneic (C57BL/6 wild-type) or allogeneic (Balb/c wild-type) heart via heterotopic cervical heart transplantation (60). Mice were monitored daily, and heartbeat of the graft was assessed via manual palpation and visualization to determine function and viability of the graft. Animals were housed under specific pathogen-free conditions under the care of the Veterinary Services Facility at Cincinnati Children’s Hospital Medical Center. All experimental procedures were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) at the Cincinnati Children’s Hospital Research Foundation.

On days 1 through 4 after transplant, mice were sacrificed. Mice were first perfused with 10mL basic salt solution (BSS) through their native heart, followed by an additional 10mL BSS perfusion of the transplanted heart. The transplanted heart was then excised, and both left and right ventricles were cut open to assess for major clots. Identified clots were then removed from the chambers as necessary, and the sample was then placed in BSS on ice. A single cell suspension was obtained from a modified cold-active protease tissue digestion protocol as previously described (61, 62). Debris from the sample was removed (Miltenyi Biotec), followed by red blood cell lysis (eBioscience). Cells were then counted and plated to prepare for staining.

Following excision of the transplanted heart, lymph nodes (including brachial, axillary, submandibular, and/or mesenteric) were identified and collected in BSS. Samples were washed and strained through a 70uM strainer to obtain a single cell suspension. Cells were then counted and plated to prepare for staining.

After blocking with 2.4G2 and Live/Dead Blue (Invitrogen) staining, cells were incubated with the GP33-PE tetramer ( Supplemental Table 1 ) for 45 minutes at 4°C. Cells were then stained with a combination cocktail of the GP276-APC tetramer along with other surface markers ( Supplemental Table 1 ) for 45 minutes at 4°C. Samples were washed and ran using spectral flow cytometry (Cytek Biosciences).

Flow cytometry analysis was performed using FlowJo (version 10.9.0). Cell populations of interest were analyzed accordingly ( Supplemental Figure 1 ). All absolute counts analyzed are depicted as ‘per heart’.

Splenocytes from two LCMV-immune mice were sorted for memory CD8⁺ T cells using a B220⁺CD11c⁺CD11b⁺CD4⁺ dump gate and then selected based on high expression of CD44. Sorted cells were plated overnight either: alone, with αCD3/CD28 Dynabeads™ (Thermo Fisher Scientific), with 5ng/mL IL-2, or with T cell-depleted splenocytes from a Balb/c mouse. T cells were depleted using the MojoSort™ Mouse CD3 Selection Kit (BioLegend). 1uL/well of the CD107a antibody ( Supplemental Table 1 ) was added to each well prior to overnight culture for assessment of degranulation. Following overnight culture, cells were stained and analyzed by flow cytometry as mentioned above.

Three immune mice receiving allogeneic heart transplants were sacrificed at days 2 (n=2) and 3 (n=1) after transplantation. Hearts were digested as described above and pool together for cell staining. The FACSymphony S6 (BD Biosciences) was used to sort for the following 6 CD8⁺ T cell populations: CD44^hiGP33⁺GFP⁺, CD44^hiGP33⁺GFP^-, CD44^hiGP276⁺GFP⁺, CD44^hiGFP276⁺GFP^-, CD44^loTetramer^-GFP⁺, CD44^loTetramer^-GFP^-. Samples with less than 3000 cells were topped off with purified B cells. These 6 samples were then processed for single-cell sequencing using the Chromium Next GEM 5’v2 Single Cell V(D)J Reagent Kit protocol. Libraries of enriched TCR sequences were created using the Single Index Kit Set A (10X Genomics) and sequenced on the NovaSeq 6000 (Illumina). Using Cell Ranger (v.6.1.2), raw files were demultiplexed and reads were aligned to the mouse reference genome package (mm10, GENCODEvM23). Loupe V(D)J (10X Genomics) was used to visualize and obtain TCR sequencing data. Sequences are deposited as NCBI Biosamples, accession numbers SAMN37531276, SAMN37531277, SAMN37531278, SAMN37531279, SAMN37531280, and SAMN37531281.

For TCR sequence analysis, we used TCRdist, an open-source Python package used for TCR sequence analysis (63), to generate sequence logo plots of the CDR3 sequences of the TCRs from the following clonotypes: CD44^loGFP^- (43 sequences), CD44^loGFP⁺ (70 sequences), GP33⁺GFP^- (66 sequences), GP33⁺GFP⁺ (21 sequences), and GP276⁺GFP^- (152 sequences). Not enough TCRαβ paired sequences were generated from GP276⁺GFP⁺ cells for analysis. For length and composition, we used the R package, alakazam (64).

Following perfusion and excision of the transplanted hearts identified for histological analysis, hearts were bisected axially to visualize the right and left ventricles and placed in 10% formalin for 24-48 hours at room temperature before moving them into 70% ethanol. Samples were then processed for paraffin embedding using the Cincinnati Children’s Pathology Research Core. Slides were cut, and hematoxylin and eosin (H&E) staining was performed to visualize pathology. Representative H&E images were taken at 10x objective magnification using a Nikon Eclipse 80i microscope with Nikon DS-Fi3 camera and NIS-elements F camera software (Nikon).

Blinded histologic scoring of H&E slides were performed using the 1990 (65) and 2004 (66) International Society of Heart and Lung Transplantation (ISHLT) guidelines. Samples were then further described using the individual characteristics making up the rejection scores ( Table 1 ).

All statistical analyses, including student t-tests, were performed using GraphPad Prism (version 10.0.0).