Longitudinal Assessment of T Cell Inhibitory Receptors in Liver Transplant Recipients and their association with Post-transplant Infections

Abstract

Current immunosuppression regimens in organ transplantation primarily inhibit T cells. However, T cells are also critical in protective immunity, especially in immune compromised patients. In this study, we examined the association of T cell dysfunction, as marked by expression of T cell exhaustion molecules, and post-transplant infections in a cohort of liver transplant patients. We focused on Programmed Death 1(PD-1) and T-cell Ig-and mucin-domain molecule3 (Tim-3) which are potent co-inhibitory receptors, and their persistent expression often leads to T cell dysfunction and compromised protective immunity. We found that patients with the highest expression of PD-1+Tim-3+ T cells in the memory compartment before transplantation had increased incidence of infections after liver transplantation, especially within the first 90 days. Longitudinal analysis in the first year showed a strong association between variability of PD-1 and Tim-3 expression by T cells and infectious episodes in transplant patients. Furthermore, T cells that expressed PD-1 and Tim-3 had a significantly reduced capacity in producing IFN-γ in vitro, and this reduced IFN-γ production could be partially reversed by blocking PD-1 and Tim-3. Interestingly, the percentage of Foxp3+ Tregs in liver transplant patients was stable in the study period. We concluded that the functional status of T cells before and after liver transplantation, as shown by PD-1 and Tim-3 expression, may be valuable in prognosis and management of post-transplant infections.

Introduction

Liver transplantation (LT) is a lifesaving procedure for end stage liver diseases. Despite excellent one year survival of greater than 89% after LT, infections continue to be one of the leading causes of morbidity and mortality in this population (1). In fact, up to 70% of patients experience at least one infectious episode during the first year after transplantation (2, 3). Nosocomial, donor derived, and opportunistic infections add to the burden early after LT, while reactivation of latent infections is common in subsequent months after transplantation (4). The increased susceptibility to infections remains even after the first 6 months due to intense immune suppression, and because of that, infection is the leading cause of death within the first 5 years of LT (1).

T cells are the principal targets of the commonly used immunosuppression drugs. However, T cells also play a central role in fighting against a plethora of pathogens. Thus, prolonged and non-specific immunosuppression after LT may compromise patients’ protective immunity, leaving patients vulnerable to infectious complications. Indeed, transplant patients experience frequent infectious episodes, and sometimes life threatening episodes (5, 6). Evidence suggests that memory T cells, both central memory and effector memory T cells, are resistant to conventional immunosuppression, and therefore capable of mediating robust recall responses to fight many bacterial and viral organisms (7, 8). However, their responses are tightly controlled by multiple regulatory mechanisms, and often acquire functionally different states, ranging from full-fledged effector cells to exhausted or dysfunctional cells (7, 9). One of the interesting aspects in this spectrum involves the opposing roles of co-inhibitory and co-stimulatory receptors in the control of effector T cell responses (10, 11). Co-inhibitory molecules inhibit T cell activation and function by competing with co-stimulatory receptor ligands and interrupting downstream signaling from T cell receptors (TCRs) (12). Sustained expression of multiple co-inhibitory receptors on T cells often marks exhausted T cells, which are frequently observed in in chronic infections and cancer (12–15).

Programmed death 1 (PD-1) receptor, a member of the CD28 family, is a prominent inhibitory molecule suppressing T cell activation pathways upon binding to its ligands PD-L1 and PD-L2 (16). Increased PD-1 expression on T cells contributes to immune dysfunction and impaired clearance of viral infections, increased nosocomial infections, and cancer (15, 17–19). T-cell Immunoglobulin and mucin-domain–containing molecule–3 (Tim-3), belonging to phosphatidylserine receptor family, is another negative regulator of T lymphocytes. Tim-3 interacts with its ligand galectin-9 and utilizes a distinct pathway to inhibit T cell responses leading to persistence of infections (20–22). Patients with viral and alcoholic hepatitis are known to have immune dysfunctions affecting T cells responses against bacteria and viruses due to increased expression of co-inhibitory receptors (9, 23–25). In those settings T cells that co-express immune inhibitory receptors PD-1 and Tim-3 become dysfunctional, with limited effector functions (14, 25).

Regulatory T (Tregs) cells are a subset of CD4+ T cells that express the transcription factor Foxp3; they have potent immunosuppressive properties and are critically involved in transplant survival (26, 27). However, the suppressive properties of Tregs can also lead to persistence of antigens and are associated with decreased clearance of chronic infections such as Hepatitis B and C (28, 29). Additionally, there is increasing evidence that inhibitory molecules such as PD-1 may play a role in development of Tregs by altering the plasticity of Th1 cells, impairing cell mediated immunity (30, 31).

In this study, we examined the dynamics of T cell co-inhibitory markers, T cell memory subsets and Tregs in peripheral blood in a cohort of liver transplant patients. In this pilot longitudinal study involving pre-transplant and post-transplant patient samples, we showed that T cells in the memory compartment with the highest expression of PD-1 and Tim-3 were dysfunctional. Importantly, sustained expression of such co-inhibitory markers post-LT was strongly associated with infectious complications.

Materials and Methods

Subjects

Peripheral blood samples were prospectively collected at regular intervals from patients who underwent orthotopic LT at the Houston Methodist Hospital Transplant Center in Houston, Texas over a fourteen-month period. After informed consent, the initial sample from each patient was obtained within 24 hours prior to LT serving as day 0 sample. Samples were subsequently collected at scheduled clinic visits post-transplantation at 1 month, 6 months and 12 months for each patient. Patients undergoing multi-organ transplant, re-transplantation within first year, patients with inadequate blood sample, patients who died within 1 year of LT and patients with samples at inconsistent time points were excluded from this study (Figure S1). Infectious episodes were defined a priori as: patients with symptoms, signs of infections with documented positive microbiologic cultures and/or viral PCR’s requiring altered management. Patients were classified into groups based on confirmed infections after LT. Blood samples from healthy and age-matched individuals were obtained from the Gulf Coast Regional Blood Center in Texas, USA.

Ethics statement

The study was approved by the Human Research Ethics Committee and Institutional Review Board (IRB # 0813-0114) at Houston Methodist Hospital and Houston Methodist Research Institute. All participants signed written informed consents form to allow collection of longitudinal blood samples for research purpose only.

Polychromatic Flow Cytometry

Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll gradient centrifugation, and stored in recovery cell culture media with 10% DMSO for phenotyping and functional experiments. The cell staining protocol and list of antibodies are described in Supplemental Materials and Methods.

Multi-color flow cytometry was performed on whole PBMCs using an LSRII instrument (BD Cytometry Systems, San Jose, CA). All data were analyzed by using Flow Jo Version 10 (Ashland, OR).

Enzyme Linked Immunosorbent assay (ELISA)

PBMCs were incubated with LEAF^™ anti-human Tim-3 antibody and LEAF^™ anti-human PD-1 antibody (10 μg/ml, BioLegend) or control IgG for an hour, followed by stimulation with anti-CD3 antibody (50 ng/ml, OKT3, Biolegend) and anti-CD28 antibody (2 μg/ml, BD Biosciences) for 48 h and 72 h. Cell supernatant was subjected to ELISA analysis for quantitative detection of IFN-γ according to manufacturer manuals (ThermoFisher Scientific, USA). The optical density at 450 nm was measured using an ELISA reader (Synergy H4; BioTek, VT, USA).

Statistics

Demographic data were reported as median and interquartile range (IQR) for continuous variables, and as frequencies and proportions for categorical variables. Demographic differences between patient groups were compared using Pearson’s Chi-square or Fisher’s exact test for categorical variables and non-parametric Kruskal Wallis test if necessary. Experimental data were analyzed using Kruskal Wallis test with post-hoc Dunn’s test for multiple comparison and Bonferroni corrections as appropriate. Wilcoxin signed-rank test was used to compare the significance of changes in IFN-γ production in PD-1 and Tim-3 blocking experiments.

Linear mixed models were used to assess the changes over time of inhibitory markers within and between the patient groups. The time factor for the models was set at day 0, 30, 180 and 365. Post hoc marginal pairwise comparisons were performed to determine the adjusted means (95% CIs) of changes in each marker of interest from day 0 to day 365. Spearman’s correlation test was performed to determine the correlation between cytokine production and inhibitory markers at each time point. Box plots and line graphs were used to depict the distribution at different time points of markers and their trend over time.

All analyses were performed using Stata version 14.2 (StataCorp LP, College Station, Texas), GraphPad Prism 7.0 (La Jolla, California), and a p-value of <0.05 was considered statistically significant.

Results

Patient enrollment

This is a single center study based on liver transplant patients at Houston Methodist over a 14-month period between January 2014 and February 2015. Sixteen patients who met the study criteria we set forth were enrolled, as well as five healthy age-matched controls (HC). A total of sixty-four PBMC samples were collected for all analyses. The enrolled LT patients were classified as those who developed infections within the first year of LT (INF) or transplant recipients who had no infections in the first year after LT (NI). Patient demographics were comparable amongst the groups (Table 1).

Table 1.

Characteristics of patients included for analyses

Characteristics	Post-transplant infection		Healthy control group (n=5)	p-value*
	Infection after LT (n=11)	No infection (n=5)
Age (years), median (IQR)	61 (51, 64)	58.4 (53, 59)	57(56, 65)	0.898
Gender, n (%)				0.80
Female	6 (54.5%)	2 (40.0%)	3 (60.0%)
Male	5 (45.5%)	3 (60.0%)	2 (40.0%)
Race, n, (%) **				0.51
White	8 (72.7%)	5 (100.0%)	NA
Hispanic	3 (27.3%)	0
Primary liver disease, n, (%)				0.43
Hepatitis C	5 (45.5%)	3 (60%)	NA
Alcoholic liver	3 (27.2%)	2 (40%)
Others ***	3 (27.2%)	0
MELD scores, **** median (IQR)	33 (32, 40)	33 (31, 35)	NA	0.375

^*Compared using the Chi-square or Fisher’s exact test for categorical variables and Kruskal-Wallis test for continuous variables,

^**Compared infection group versus non-infection,

^***Others – polycystic liver disease and sub fulminant liver failure,

^****Compared using Mann Whitney U test

Pre-and post-transplant laboratory parameters at various time points for each patient were monitored by the treating physicians at follow up visits and the median values were comparable between the two patient groups (Table 2). The standard immunosuppression regimen immediate post-LT in our center included Prednisone, Tacrolimus, and Mycophenolate Mofetil (MMF). Tacrolimus was initiated within the first three days after transplantation at a dose of 1–2 mg/kg body weight/day. Drug levels were monitored closely and dose adjusted to maintain adequate concentrations. The median Tacrolimus trough level was different at day 180 between patient groups, but no difference at other time points. No difference was noted in the steroid regimens and MMF doses between the groups (Table S1).

Table 2.

Details of infectious complications of patients

Patient no.	Age	Sex	# IE in first year post-LT	Time of IE since LT (days)	Organisms detected	Symptoms and signs	Management
1	65	F	2	58	Enterovirus pneumonia	Acute respiratory distress, fevers and cough, lymphocytosis	Hospital admission, respiratory support. LOS-7 days
				274	Clostridium difficile, Varicella zoster	Bloody diarrhea, lymphocytosis, shingles	Hospital admission, Metronidazole, Acylcovir, LOS-7 days
2	36	F	3	58	Corona virus pneumonia	Fever, productive cough, CXR – R upper lobe consolidation	Hospital admission, respiratory support. LOS-4 days
				247	Bilateral MRSA thigh abscesses	Fevers, draining wounds on both thighs, leucocytosis	Hospital admission, Vancomycin. LOS-7 days
3	61	M	1	31	Enterococcus cellulitis	Draining groin wound, abscess	Hospital admission, wound drain and fascia repair, LOS-4 days
4	59	F	1	39	E. Coli Urosepsis	Hypothermia, tachycardia	IV Ciprofloxacin
5	65	M	1	88	Influenza A, Enterovirus pneumonia	Cough, CXR-R lower lobe infiltrate with pleural effusion	Tamiflu, Respiratory support LOS-5 days
6	49	M	2	32	Pseudomonas Urosepsis	Leucocytosis, fevers	Ceftazidime
				366	Varicella Zoster	Fevers, Shingles, PCR positive	Acyclovir, LOS-6days
7	62	M	1	68	Adenovirus diarrhea, Clostridium difficile	Bloody diarrhea, leukopenia	Metronidazole
8	63	M	1	11	Enterococcus fecalis cellulitis left thigh	Left thigh swelling, drainage	Incision drainage, Vancomycin, LOS-3 days
9	72	F	1	116	MRSA cellulitis of knee	Leg and knee swelling, suprapatellar effusion	Drainage, Vancomycin, LOS-12 days
10	51	F	2	88	E. coli UTI	Right flank pain, hypothermia	IV Ciprofloxacin
				199	E. coli sepsis	Fever, nausea, abdominal pain	Piperacillin/Tazobactam, LOS 32 days
11	50	F	1	86	Klebsiella UTI	Fever, vomiting	Imipenem/Cilastatin LOS 5 days

IE – infectious episode, CXR – chest X ray. LOS-length of hospital stay. 3 patients in INF group and 1 patient in NI group had rejection episodes requiring increased IS management between 6–9 months post-LT

Patients presenting with signs and symptoms of infections underwent evaluation and management by treating physicians. A majority of patients presented with initial episode of infections within the first 3 months post -LT (Table S2). Adjustments of immunosuppressive drugs were made by treating physicians if necessary, and normal levels of immunosuppression were applied again after the clearance of infections.

Fourteen out of 16 patients were CMV IgG positive pre-transplantation. Patients were on prophylaxis for CMV per protocol after LT and monitored for CMV-DNA by PCR. None of the patients in our study cohort developed CMV viremia after LT.

Characteristics of memory T cell subsets after liver transplantation

We prospectively collected PBMC samples in a longitudinal fashion from patients undergoing liver transplant at set time points. The initial sample was obtained pre-transplantation (24 hours prior to the patient undergoing transplant), after LT at 1 month, 6 months and 12 months. Peripheral T cell subsets were analyzed in PBMCs to distinguish naïve from memory cells by polychromatic flow cytometry on the basis of CCR7 and CD45RO expression by CD4⁺ and CD8⁺ T cells (Figure 1A) (33). Patient groups were compared to age matched healthy controls (n=5).

Figure 1. Distribution of circulating CD4⁺ and CD8⁺ memory T cell subsets after LT.

(A) Gating strategy of PBMC from liver transplant patients is shown. Lymphocytes were identified by scatter properties and viable CD3+ lymphocytes were divided into CD4+ and CD8+ T cell subsets.

(B and D) Representative flow plot shows gating strategy to identify CD8 memory subsets (Fig 1B) and CD4 memory subsets (Fig 1D). Central Memory (CCR7+CD45RO+, CM) and Effector memory (CCR7-CD45RO+) percentages were compared between pre-transplant data points of LT patients and healthy controls (HC, n=5) on CD8 T cells (Fig 1B) and on CD4 T cells (fig 1D). LT patients (n=16) were divided into INF (n =11, patients who developed infections after LT) and NI (n=5, patients without infections after LT). Horizontal lines indicate median values and IQR (Kruskal Wallis test). CM, central memory; EM, effector memory.

(C and E) Proportion of memory subsets on CD8 (Fig C) and CD4 (Fig E) were computed for each patient over time based on CCR7 and CD45RO at four time points. We compared pre-transplant baseline sample of each patient and post LT samples obtained at 30 days, 180 days and 365 days using linear mixed models. Box plots are showing median ± IQR *p<0.05 was considered statistically significant. CM, central memory; EM, effector memory.

As shown in Fig 1B, effector memory T cells or T_EM (CCR7⁻ CD45RO⁺) were the predominant subset among CD8+ T cells. In the CD8 compartment, the pattern of memory T cell subsets was similar in patient subgroups and healthy controls pre-transplantation (Figure 1B). The ratio of T_EM to T_CM was 3.5:1 in the CD8+ T cells in patients pre-LT. There was an increased proportion of T_CM after LT (total median proportions for all patients, n=16 - pre-LT 7.4%, 1 month 15%, 6 months 15.6% and 12 months 11.1%; median values). When the patient groups were compared longitudinally, the proportional distribution of the memory subsets remained stable with no difference between the groups (Figure 1C).

In CD4+ T cells, the overall median proportion of T_CM was 43% and T_EM was 9.8% pre-LT (n=16), with central memory being the dominant memory subset at all time points. When the baseline proportions of the two patient groups (Infection, INF and no infection, NI) compared to healthy controls (HC), the distribution of CD4+ T_CM was similar (Figure 1D). On longitudinal analysis of patient groups, there was no difference in the distribution of T_CM over time after transplantation (Figure 1E). However, NI had significantly higher median proportions of CD4 T_EM compared to INF at the baseline pre-LT time point, but there were no differences over time after transplantation.

Overall, the proportion and distribution of central versus effector memory T cells were maintained after liver transplantation in both the CD8 and the CD4 compartments.

Pre–transplant expression of PD-1 and Tim-3 is associated with infectious episodes after LT

We compared the expression of co-inhibitory molecules on T cell memory subsets pre transplantation in patients undergoing LT. We quantified the percentages of T cells expressing co-inhibitors PD-1 and Tim-3 (double positive cells, DP cells) on CD8+ and CD4+ memory T cells from patients at baseline (day 0) and compared that with healthy controls.

Representative graphs of PD-1⁺Tim3⁺, DP cells in CD8 T_EM and CD8 T_CM in patients with infections and without infections after LT were shown (Figure 2A). Patients who developed infections (INF) after LT had significantly higher expression of the co-inhibitors pre-transplantation in the CD8+ memory subsets, most strikingly in the CD8 T_EM subset (Figure 2B), the median percentage of DP cells was 16.4% in INF group, as compared to 1.1% in NI group. The percentage of DP cells in healthy controls was ~1 % (overall p=0.008). Similar trends were noted in CD8 Tc_M subset as well, with median 1.7% DP cells at baseline in INF group compared to 0.5% and 0.2% in NI and HC, respectively (p=0.036).

Figure 2. Pre-transplant high PD-1 and Tim-3 co-expression on memory T cell subsets is associated with infections after LT.

(A) Flow cytometry plot showing expression of DP cells of PD1+ and Tim-3+ on CD3+CD8+ memory subsets. Representative graphs on the left panel is from a pre-transplant sample of a patient who developed infection after LT and on the right, is a patient without infections in the first year after LT. Top panel depicts co-inhibitor expression (PD1+Tim3+) pre-transplant on CD8 central memory and bottom panel on CD8 effector memory.

(B) Comparisons of 3 groups for baseline expression of co-inhibitors PD1 and Tim3 on CD8 effector memory and central memory. Percentage of PD1 and Tim3 DP cells on the memory T cells subsets were analyzed. The pre-transplant values of PD1+Tim3+ for both patient groups (INF-infection n=11, NI - no infection after LT n=5) were considered baseline and compared with healthy control (n=5). The red dots represent patients who developed infections after LT, blue squares showing patients who had no infections in the first year after LT and the green triangles represent healthy controls.

(C) Similar expression of PD1+Tim3+ on CD4 memory subsets CM and EM were obtained. Plot shows the comparison of median percentages between the three groups with IQR (Kruskal Wallis test) p<0.05 considered significant

We also studied the differences in CD4+ T cell memory subsets amongst groups. The median percentages of PD-1⁺ Tim-3⁺ DP cells in CD4+ T cells were compared between the 2 patient groups and HC (Figure 2C). In CD4+ T_EM, the median percentage of DP cells was 3.6% in INF and 1.6% in NI. The baseline expression in HC was 1.1% in effector memory and 0.4% in central memory CD4+ T cells (p=0.036 and p=0.027 respectively).

Taken together, we found the highest percentage of PD-1 and Tim-3 co-inhibitory receptor expression in T_EM in the CD8 subset in patients pre-transplantation who subsequently developed infections after transplant. This upregulation was observed in patients who developed viral and bacterial infections post-transplant, with a median percentage expression of DP cells being 20.2% and 14.7% respectively (Figure S2).

Patients with increased expression of PD-1 and Tim-3 after LT experience increased infectious episodes

We did longitudinal assessments of PD-1 and Tim-3 expression over time after transplant to examine the progression of these markers in relation to transplantation. A total of 4 samples were collected from each patient including the pre-transplant blood sample. We examined percentages of PD-1 and Tim-3 over time in the first year after LT and analyzed using linear mixed models on both CD4+ and CD8+ T cells. In CD8+ T_EM cells PD-1⁺Tim-3⁺ cells continued to be significantly higher in the INF group than the NI group in the first year after LT. There was no marked difference in PD-1 and Tim-3 expression between the patient groups by day 365 (Figure 3A). The overall median expression of the co-inhibitors was low by day 365, which was ~3% in both groups. Of note, most infections occurred in the first 3–6 months after LT in the patient cohort. The differences were less pronounced in the CD8+ T_CM subset with higher PD-1 and Tim-3 expression in INF group at day 30 and day 180, but the differences in groups did not achieve statistical significance. In this longitudinal analysis, the co-inhibitor expression was higher at day 30 after LT both on CD4+ central and effector memory cells (Figure 3B).

Figure 3. High and dynamic PD-1 and Tim-3 expression on memory T cell subsets in patients with infections after LT.

(A) The frequencies of co-inhibitor expression over time on CD8 T_EM and CD8 T_CM were analyzed by flow cytometric determination based on expression of both PD1 and Tim3. We compared pre-transplant baseline sample of each patient and post LT samples obtained at 30 days, 180 days and 365 days. Panel on the left indicates effector memory subset and on right central memory determined based on CCR7 and CD45RO measurements. Horizontal line in box plot indicates median values. Box plots are showing median ± IQR, red box plots represent patients who had infections after LT and blue plots, patients without infections in first year after LT. *p<0.05 was considered statistically significant.

(B) Analysis over time of CD4 memory subsets for PD1 and Tim3 co expression was calculated. All four samples from a patient analyzed on a single day by flow cytometric analysis. Left panel indicates effector memory subset and right panel central memory determined based on CCR7 and CD45RO measurements Graphs depict Median with IQR obtained by linear mixed models. * p <0.05 considered statistically significant

We constructed trend lines to show PD-1 and Tim-3 DP cells in each patient over time after LT (Figure 4). Patients who had no infections in the first year after LT had uniformly low levels of PD-1 and Tim-3 expression, which was under 5% and was stable, exhibiting little variability over time, on both CD4+ and CD8+ memory subsets. In contrast, patients with infections had elevated and highly variable expression of PD-1 and Tim-3 after LT, particularly on the effector memory subsets. Trend line representations on CD8+ T_EM were shown in Fig 4A. Similar variability was seen in the CD4+ T_EM subset in the INF group (Fig 4B). The median percentage of PD-1 and Tim-3 DP cells in HC was under 1% in the memory subsets of CD4+ and CD8+ T cells, respectively. Altogether, these data indicate that patients who had infections after LT had elevated and highly variable expression of PD-1 and Tim-3 by CD4+ and CD8+ T_EM cells after liver transplantation.

Figure 4. PD-1 and Tim-3 expression is a dynamic process after LT.

(A) Trend lines of each individual patient plotted over time for PD1+Tim3+ on CD8 effector memory. Top graph shows patients with infection and bottom plot patients without infection after LT. The variability and dynamic nature of these inhibitor markers in infection group is striking.

(B) PD1+Tim3+ expression on CD4 effector memory was plotted to demonstrate trend over time of the inhibitory markers over time. Patients with infection had variability and notably high expression even at pre-transplant time point. The expression of PD1 Tim3 (DP cells) was uniformly less than 5% at all time points in patients who did not develop any infections in the first year after LT.

Cytokine profiles in patients with high PD-1 and Tim-3 expression after LT

We assessed the functional status of CD4+ and CD8+ T cell memory subsets in LT patients by staining for expression of effector cytokines, as well as the cytolytic marker CD107α. PBMCs from patients were rested overnight, briefly stimulated with PMA/Ionomycin, then stained for memory markers, followed by intracellular staining for cytokines. Healthy control cells processed under identical conditions or unstimulated PBMCs for each patient were included as controls.

As shown in Fig 5, the percentage of IFN-γ positive T cells after in vitro stimulation was higher in patients who had no infections, as compared to patients who had infections after LT. At day 30 post-transplantation, there were higher percentages IFN-γ producing CD8+ effector memory and central memory T cells in NI group (Fig S3A, red box -infection after LT and blue box no infections after LT). In contrast, the percentage of TNF-α positive cells were lower at day 30 after transplantation in the CD8+ T_CM subset, but no differences were found at other time points (Fig S3B). CD107α expression was similar between the patient groups at all time points examined (Fig S3C).

Figure 5. Inverse correlation of IFN-γ production with PD-1 and Tim-3 expression in LT patients.

(A) PBMCs were stimulated in vitro for 5 hours with PMA and Ionomycin to determine cytokine production. Intracellular staining performed for IFN-γ and TNF-α and acquired on flow cytometer. Unstimulated stained cells served as negative controls.

(B) Association between cytokine and co-expression of PD-1 and Tim-3 on CD8 effector memory and (C) CD4 effector memory is shown. Negative correlation of co-inhibitor expression with IFN-γ but not with TNF-α. Spearman r and p value of correlations are depicted in the upper right corner of each graph. Lines represent linear regression. EM – effector memory

The percentage of IFN-γ producing CD4+ T cells in effector memory subset at day 30 was lower in the INF group when compared to NI (Fig S4A). However, there was no difference in the percentage of TNF-α positive CD4 T cells between groups on both effector and central memory subsets at various time points after LT (Fig S4B).

We further measured the correlation between expressions of PD-1 and Tim-3 and cytokine production. As noted above, the percentage of IFN-γ positive T cells was lower in patients who developed infections. However, these patients expressed higher percentages of PD-1 and Tim-3 DP cells in the CD8+ and CD4+ effector memory subsets. Thus, there was a significant negative correlation between expression of PD-1 and Tim-3 on CD8+ T_EM cells and IFN-γ production (rho −0.30, p 0.017), and this negative correlation in CD4+ T_EM subset did not reach significant level (rho −0.23, p 0.07) (Fig 5B and C). TNF-α expression in PD-1 and Tim-3 DP cells in either CD4+ or CD8+ T_EM cells did not show significant correlation either.

Antibody-mediated PD-1 and Tim-3 blockade partially restored IFN-γ production in vitro

To further address the functional role of PD-1 and Tim-3 in T cell exhaustion in liver transplant patients, PBMCs from patients with infectious episodes were stimulated with OKT3 mAb plus anti-CD28, and in some cultures we added both anti-PD-1 and anti-Tim-3 mAbs. We examined the IFN-γ production 48 and 72 hrs after the culture by ELISA assays. PBMCs activated in the presence of isotype control Abs were included as controls. Consistent with previous studies (14, 34–37), ELISA assays of cell supernatants showed that concentration of IFN-γ was significantly increased in the PD-1 and Tim-3 blocking groups when compared to the Isotype control (Figure 6A). The IFN-γ secretion was higher at 72 hours as compared to 48 hours with median values of 10,939 pg/ml and 6776 pg/ml respectively (Figure 6B). These results suggest that PD-1 and Tim-3 inhibitory pathways suppress IFN-γ production and dampen the T cell effector responses.

Figure 6. Role of PD-1 and Tim-3 blockade in IFN-γ production in vitro.

PBMCs from 10 representative samples were incubated with anti-PD1 and anti-Tim-3 antibodies. Duplicates were incubated with Isotype control. Cells were stimulated with OKT3 and anti-CD28 for 48 hours and 72 hours and cell supernatant collected for ELISA analysis of IFN-γ. Blocking group had significantly higher concentration of IFN-γ (pg/ml) both at 48 hours (Fig 6A) and at 72 hours (Fig 6B) with continued increased levels at 72 hours. Lines represent paired data.

Characteristics of regulatory T cell (Tregs) in liver transplant patients

CD4⁺CD25⁺Foxp3 (Tregs) have emerged as key immunosuppressive cells in the immune system. To examine the Treg population in peripheral circulation in the patients, we stained PBMCs with CD4, CD25, as well as Foxp3 antibodies. The median percentage of Tregs at baseline in LT patients (n=16) was 3.6%, which was similar to healthy controls at 4% (Fig 7A). In the liver transplant patients, we measured the percentages of Tregs over time. In our longitudinal assessments, patients with infections (INF) exhibited higher percentages of baseline Tregs before transplantation when compared to patients without infections (NI), but after transplantation, we did not find statistically significant differences between the patient groups at all time points examined (Fig 7B).

Figure 7. Regulatory T cell percentages remain stable after LT.

CD4+CD25+Foxp3 regulatory T cells (Tregs) percentages were determined for each patient and healthy control on unstimulated stained cells. (A) The patient groups were compared to healthy control in a Kruskal Wallis tests. Red dots showing LT patients with infections (n=11), blue squares LT patients without infections (n=5) and green triangles healthy controls (n=5). Median and IQR percentages are shown in dot plots. (B) Tregs percentages of patient groups were compared over time after transplant. Red box plots represent patients who had infections and blue plots patients without infection.

Discussion

In the present study, we focused on a cohort of liver transplant patients and examined the expression of T cell co-inhibitory molecules in relation to infectious episodes after LT. We demonstrated that patients with high expression of the co-inhibitory molecules PD-1 and Tim-3 before transplantation had a strong tendency to develop infections after LT. Interestingly, concurrent PD-1 and Tim-3 expression was particularly prominent on the memory CD8+ T cells, especially on the CD8+ effector memory subset; such patients frequently developed infections within the first 90 days after transplantation. Furthermore, we provided data, based on longitudinal assessments of PD-1 and Tim-3 expression over time that increased post-transplant expression of PD-1 and Tim-3 by effector memory cells (both CD8+ and CD4+ T cells) was correlated with increased incidence of infections within the first year after transplantation. We observed that patients without infections in the first year after LT had substantially low levels of PD-1 and Tim-3 expression on either T cell subsets both pre- and post-liver transplantation.

T cells play a pivotal role in graft rejection as well as in eliminating infections, which presents an interesting paradox in the management of transplant patients. Thus far, the underlying mechanisms for the incidence of post-transplant infections after liver transplantation are unknown. Certain demographic factors such as advanced age, malnutrition, and high MELD scores have been suggested as contributing factors in post-transplant infections (38–40). There is a growing appreciation that T cell exhaustion due to sustained expression of T cell co-inhibitory molecules often results in compromised protective immunity, and consequently, chronic and persistent infections usually occur (15, 17, 25). In fact, T cell exhaustion was initially described in chronic choriomeningitis mouse model, and more recently also in HIV, cancer, hepatitis C, bacterial sepsis and viral infections (36, 41–43). Though exhaustion of donor specific T effector cells may be beneficial to graft survival, we provide evidence that in liver transplant patients, high expression of PD-1 and Tim-3, which functionally mark dysfunctional T cells, is associated with infectious episodes following transplantation. One outstanding feature in transplantation is that patients must take immunosuppression drugs for life, which broadly suppress the immune responses, including responses to pathogens. But memory T cells are shown less susceptible to such drugs as compared to their naïve counterparts; they are also resistant to depletion therapies, persisting in the circulation long after broad T cell depletion treatments (33). Thus, it is conceivable that memory T cells may play a particular important role in fending pathogens in transplant patients, and their dysfunctions may render patients particularly vulnerable to infections (48). Indeed, in our patients the most striking changes regarding PD-1 and Tim-3 expression were found in the memory compartment where concurrent expression of both PD-1 and Tim-3, either pre- or post-LT, predisposes patients to infections. In fact, some patients in our cohort in the INF group had greater than 30% of CD8 T_EM cells expressing both co-inhibitory receptors, a feature that was not observed in NI group or the healthy controls. In other models, increased co-inhibitory markers on memory CD8+ T cells are also associated with a dysfunctional state, resulting in inability to clear pathogens and development of chronic diseases (16, 42, 44–46).

What caused the high and sustained expression of PD-1 and Tim-3 on T cells, especially in patients before transplantation, was not examined in the present study, and therefore remains unknown. But the primary liver diseases (e.g., hepatitis C infections), or history of infections, medications and vaccinations may be contributing factors. However, some liver transplant patients in our studies who were transplanted due to hepatitis C infections exhibited low levels of PD-1 and Tim-3 expression and did not experience infectious episodes. Such co-inhibitory receptors can be expressed by recently activated T effector cells, but all patients were screened and did not show features of infections before LT. We suspect that for those who experienced infectious episodes after transplantation but also showed high expression of PD-1 and Tim-3 on memory T cells, some of the PD-1+Tim-3+ T cells might be recently activated T cells responding to pathogens, and such T cells may not necessarily be exhausted T cells. Along the same line, the question that in those patients, the exceeding high proportion of PD-1+Tim-3+ T cells in the memory compartment is the cause or consequence of infections is difficult to discern in our current study. Our data did highlight a strong association between such two events. The finding that T cells from patients who developed infections had impaired IFN-γ expression, even after in vitro stimulation, and that blockade of both PD-1 and Tim-3 could restore IFN-γ production does suggest a substantially compromised effector response in those patients.

The Foxp3+ T regulatory cells (Tregs) have gained tremendous interest in the field of transplantation due to their immune suppressive properties required for the induction of allograft tolerance (47). In our study, we failed to observe marked changes in the relative percentages or the numbers of Tregs in PBMCs, and the overall Tregs were comparable between the patient groups and healthy controls. However, patients who developed infections had a significantly higher percentage of Tregs pre-transplantation, and the reason behind this increase was also unknown. On longitudinal assessment, the Tregs percentages remained similar and stable between patient groups after-transplantation regardless of infectious episodes. Conceptually, an expanded Treg pool, along with overexpression of PD-1 and Tim-3 by memory T cells or a high proportion of exhausted T effector cells, may further hinder protective immune responses, a situation that can be exacerbated in transplant patients by additional immunosuppression drugs. This is a significant issue that warrants further investigation involving a bigger cohort of patients.

Our data may have important clinical implications in management of liver transplant patients. The association of PD-1 and Tim-3 expression, which is easily accessible and measurable, with infectious episodes both before and after transplant, is of potential diagnostic or prognostic values in the clinic. Currently, most institutions manage transplant patients with broad immune suppression protocols, which are associated with increased incidence of infections in the first year after transplantation (4). Our data suggest that an early assessment of markers of T cell exhaustion pre-transplantation may help identify patients at-risk for infections, so that immunosuppression treatments can be adjusted and management of potential infections be considered. Also, measurements of T cell inhibitory markers post-transplantation may be useful in predicting risks of infections following liver transplantation. This personalized approach according to patient’s immune status is an emerging area in management of morbidity and mortality in susceptible patients.

This study also has limitations including a small sample size and challenges in obtaining adequate quantity of blood lymphocytes consistently from immunosuppressed patients at all time points. The patient samples are also biased toward certain racial groups in our study. Hence, it remains unknown if race has any influence on expression of these markers. Though the immunosuppression regimen was uniform amongst all patients during the study peroid, the Tacrolimus trough levels was different at day 180 between the infection and non-infection group, whether this would impact the outcomes requires clarifications in future studies. But most of the infections occurred during the first three months after transplantation. In addition, there are other cell surface markers that are associated with T cell exhaustion, and their expression pattern(s) in liver transplant patients warrants further investigation. Nevertheless, considering the strong association we identified and the negative impact of infections on liver transplant outcomes, further studies based on multi-center and longitudinal settings will be required in moving the field forward.

Supplementary Material

Supp FigS1-4

Figure S1: Patient selection criteria.

Figure S2: Pre-transplant PD-1 and Tim-3 expression on CD8+ T effector memory cells in association with post-LT bacterial and viral infections.

Figure S3: Longitudinal assessment of cytokine production by CD8+ memory T cell subsets.

Figure S4: Cytokine production on CD4 memory subsets is shown.

Supp TableS1-3

Table S1: Immunosuppression regimen

Table S2: Pre and post-LT laboratory parameters at each time point

Table S3: Stratifying infections by viral and bacterial etiologies

Abbreviations

AR

acute rejection

CMV

cytomegalovirus

LT

liver transplantation

MMF

Mycophenolate Mofetil

PBMC

peripheral blood mononuclear cells

TCM

central memory T cells

TEM

effector memory T cells

T Naive

naıve T cells

, PD1

Programmed death1

Tim3

T cell Immunoglobulin and mucin-domain–containing molecule–3

Tregs

Regulatory T cells

DP

double positive cells