The Fractalkine Receptor CX3CR1 Links Lymphocyte Kinetics in CMV-Seropositive Patients and Acute Myocardial Infarction With Adverse Left Ventricular Remodeling

Aims Latent cytomegalovirus (CMV) infection is associated with adverse cardiovascular outcomes. Virus-specific CX3CR1+ effector memory T-cells may be instrumental in this process due to their pro-inflammatory properties. We investigated the role of CX3CR1 (fractalkine receptor) in CMV-related lymphocyte kinetics and cardiac remodeling in patients with ST-elevation myocardial infarction (STEMI) undergoing primary percutaneous coronary intervention (pPCI). Methods and Results We retrospectively analysed lymphocyte count, troponin, and survival in 4874 STEMI/pPCI patients, evaluated lymphocyte kinetics during reperfusion in a prospective cohort, and obtained sequential cardiac MRI (cMRI) to assess remodeling. Pre-reperfusion lymphopenia independently predicted mortality at 7.5 years. Prior to reperfusion, CCR7+ T-lymphocytes appeared to be depleted. After reperfusion, T-lymphocytes expressing CX3CR1 were depleted predominantly in CMV-seropositive patients. During ischaemia/reperfusion, a drop in CX3CR1+ T-lymphocytes was significantly linked with microvascular obstruction in CMV+ patients, suggesting increased fractalkine-receptor interaction. At 12 weeks, CMV+ patients displayed adverse LV remodeling. Conclusion We show that lymphopenia occurs before and after reperfusion in STEMI by different mechanisms and predicts long-term outcome. In CMV+ patients, increased fractalkine induction and sequestration of CX3CR1+ T-cells may contribute to adverse remodeling, suggesting a pro-inflammatory pathomechanism which presents a novel therapeutic target.


INTRODUCTION
The immunological component of atherosclerosis and coronary artery disease has garnered increasing interest in the last decade. Recent phase III trials of anti-inflammatory agents in humans show promising, although inconsistent results (1)(2)(3). The CANTOS trial demonstrated the potential for targeting specific pathways in the innate immune system (1), but similarly targeted inhibition of the adaptive immune system has not yet yielded such favourable results (4).
Having previously shown that T-lymphocytes are crucial mediators of the response to myocardial ischaemia and reperfusion (5), we are now the first to analyse how this response differs in the CMV-acquainted lymphocyte compartment, and how it affects remodeling after myocardial infarction.

METHODS
Details of the major resources and detailed methods can be found in the online-only Data Supplement.

Patient Populations
This study utilises several different cohorts of patients. We retrospectively analysed 4874 consecutive STEMI/pPCI patients from our centre between 2008 and 2015. We analysed lymphocytes and CX 3 CR1 in whole blood throughout reperfusion in 52 STEMI/pPCI patients recruited for CAPRI (Ciclosporin to Reduce Reperfusion Injury in Primary PCI Clinical Trial) (14) (demographics in Table 1), and analysed three further prospective cohorts of STEMI/pPCI patients, and 10 healthy controls. These populations are shown schematically in Figure 1.

CMV Serostatus
CMV serostatus in all cohorts was determined with Roche Elecsys CMV IgG immunoassay. CMV seropositivity was defined as an IgG index >1.0 U/mL, as per manufacturer's instructions.

Cardiac Magnetic Resonance Imaging
In CAPRI, acute cMRI was obtained at 2-7 days post-reperfusion in 51 patients, and repeat cMRI at 12 weeks was obtained in 48 patients to assess remodeling. In another study 50 cMRI scans were acquired 1-8 days post-reperfusion (5), providing a total of 101 acute cMRIs. Parameters of interest were infarct size as a percentage of the left ventricle (LV), LV ejection fraction, enddiastolic LV volume, end-systolic LV volume and microvascular obstruction (MVO), and were measured as previously described and validated (5).

Immunohistochemistry of Human Hearts
Left ventricular myocardial tissue samples were obtained from failing and non-failing human heart explants (n=21), the latter that were not accepted for transplantation. Eight were female, mean age was 62 ± 12 years, mean left ventricular ejection fraction 64 ± 7%. Tissue samples were collected at the time of explantation, formalin fixed, and paraffin embedded. Immunohistochemistry was performed using anti-human CD3 primary antibody (Dako, G A503) according to manufacturer's recommendations. Quantification was performed manually by a pathologist blinded to CMV serostatus.   Table 4). Only variables that were independently associated with mortality by multivariate Coxregression were selected to compose the final prediction model. Univariate associations of lymphocytes with overall mortality were assessed by Kaplan Meier survival analysis and unadjusted Cox-regression. In the prospective cohort, although no baseline characteristics were significantly associated with CMV serostatus, those which appeared clinically relevant and differed between CMV serogroups were selected for multivariate linear regression analysis of change in end-diastolic volume ( Table 2) and endsystolic volume (Supplemental Table 7).

CMV Seropositive Patients Show Adverse Left-Ventricular Remodeling After STEMI
We first investigated whether CMV serostatus affected remodeling of the left ventricle in the first 12 weeks after acute STEMI/pPCI ( Figure 2A). By comparing 101 cMRIs within 8 days of STEMI/ pPCI (58 CMV+, 43 CMV-; demographics in Supplemental  significant deterioration in end-diastolic volume (+10.7ml vs -6.1ml; p=0.02), but no significant difference in change in endsystolic volume (-2.0ml vs -9.1ml; p=0.27). The effect of CMV serostatus on LV remodeling increased after multivariate analysis adjusting for age, smoking status, statin usage, history of hypercholesterolaemia, peak high-sensitivity troponin T and infarct location (anterior vs non-anterior). The adjusted change in EDV in CMV seropositive patients compared to seronegative patients was +22.5mL (95% CI 7.0 -37.0mL; p=0.005; Table 2), and for ESV was +13.6mL (95% CI -1.0 -28.1; p=0.068;  Three of the four prospective cohorts were recruited between 2008 and 2015, so these patients are also included in the retrospective analysis. The TACTIC trial recruited patients after 2015, so these are not included in the retrospective cohort. cMRIs were unanalysable either because of technical problems with the scanner or poor image quality (e.g. due to atrial fibrillation). Acute cMRIs from Boag et al. and CAPRI were analysed together. We also describe data from 10 healthy control patients, and 21 explanted human hearts, not shown in this diagram.  Table 7). We hypothesised that CMV-related differences in T-lymphocytes drive inflammation after STEMI, leading to adverse remodeling at 12 weeks.

Early Post-Reperfusion Loss of CCR7-Negative T-Lymphocytes Is Associated With Microvascular Obstruction in CMV Seropositive Patients
Coronary MVO is an independent predictor of adverse outcomes after STEMI (16), and we have previously shown an association between MVO and early post-reperfusion loss of effector (CCR7 -) T-lymphocytes, with cell depletion 15-30 minutes after reperfusion most strongly predictive of MVO (5). We show here that this association is exclusively due to CMV seropositive patients ( Figure  2B). To account for lower pre-reperfusion counts of effector Tlymphocytes in CMV seronegative patients, we compared the percentage change in cell numbers, rather than absolute, in patients with no, low and high MVO. CD4 + T EMRA are effectively absent in CMV seronegative patients ( Figure 3A) and so we expected to see no changes in this cell population, but they are included in the analysis for completeness. In CMV seropositive patients, the high-MVO group showed a significantly greater percentage drop than the zero-MVO group in all effector memory subsets: CD4 + T EM (-23% vs -5%; p=0.0001), CD4 + T EMRA (-34% vs -4%; p=0.0002), CD8 + T EM (-26% vs -12.5%; p=0.011) and CD8 + T EMRA (-35% vs -15%; p=0.021). In CMV seronegative patients, despite a non-significant trend towards greater lymphocyte drop in patients with high MVO than low MVO, there was no significant differences in lymphocyte count changes between the zero-, lowand high-MVO groups despite similar sample sizes in CMV serogroups. CCR7 + T-lymphocytes were not associated with MVO in either serogroup.

CX 3 CR1-Positive T-Lymphocytes Drop Acutely After Reperfusion in CMV Seropositive Patients
Having shown that dynamic changes in effector T-lymphocyte populations after reperfusion predict adverse cMRI findings in CMV seropositive patients, we aimed to characterise these changes in detail ( Figure 3A).   reperfusion. In contrast, CD4 + T EM , CD4 + T EMRA , CD8 + T EM and CD8 + T EMRA , all of which are CCR7effector memory cells, behaved differently in CMV seropositive and seronegative patients. In CMV seronegative patients they resembled CCR7 + cells, with minimal change in response to reperfusion. In CMV seropositive patients, however, they dropped significantly over the first 90 minutes after reperfusion, returning to near pre-reperfusion levels at 24 hours. At 90 minutes, CMV seropositive patients had lost almost all the excess CD8 + T EM that they had prior to reperfusion. This suggests that in CMV seropositive patients, CCR7effector memory T-lymphocytes do not fall in response to ischaemia (as they were no lower before reperfusion than at 24 hours), but rather they fall in response to reperfusion. Subsets which drop significantly after reperfusion all express CX 3 CR1 ( Figure 3B), confirming this as a possible mediator of post-reperfusion lymphocyte depletion in CMVseropositive patients.

Pre-Reperfusion Lymphopenia in STEMI Patients Correlates With Higher Troponin T Values, and Predicts Long-Term Mortality
We have detailed the response of lymphocytes to reperfusion, but assessing dynamic changes due to myocardial infarction itself is more challenging, with pre-ischaemic values rarely available. To investigate whether lymphocytes fall in response to myocardial ischaemia, prior to reperfusion, we analysed the relationship between lymphocyte count and high-sensitivity troponin T (hsTnT) on admission blood samples in 4874 consecutive STEMI patients undergoing pPCI at our center (Supplemental Table 3). At admission, prior to pPCI, lower lymphocyte counts correlated with a higher number of anterior infarcts (p<0.001) and higher hsTnT ( Figure 4B, p<0.001). This suggests that blood lymphocyte counts fall during myocardial infarction, to an extent dependent on the size of the infarct and length of ischaemic time, prior to coronary intervention. Earlier work by our group has shown that lymphopenia one day following reperfusion predicts  long-term mortality (5), and here we found that patients with lower lymphocyte counts on admission, prior to reperfusion, also had significantly worse long term survival (hazard ratio for lowest vs highest quartile after adjusting for all covariates: 1.37; 95% CI 1.1-1.7; p=0.004; Supplemental Table 4 and Figure 4A). This divergence in survival curves was evident within a few months and increased throughout the 7.5-year mean followup period.

Pre-Reperfusion Leukocyte Count Correlates With Infarct Size and Left-Ventricular Ejection Fraction in CMV Seronegative Patients
Survival after STEMI is largely determined by left-ventricular ejection fraction (LVEF) (17), so we investigated the relationship between pre-reperfusion lymphocytes and LVEF in 52 STEMI patients (demographics in Table 1). Lower pre-reperfusion total lymphocyte count was correlated with lower LVEF at 2-7 days after STEMI, but only in CMV seronegative patients (r=0.839, p<0.001; Figure 4C). CMV seropositive patients showed no correlation between admission total lymphocyte count and LVEF (r=0.017). We suggest that this relationship is due to lymphocytes falling in larger infarcts, which lead to worse LVEF but only in CMV seronegative patients. Interestingly, CCR7 + T-lymphocyte subsets all showed significant correlation between pre-reperfusion count and LVEF, again only in CMV seronegative patients (CD4 + T N : r=0.530; CD4 + T CM : r=0.502; CD8 + T N : r=0.689; CD8 + T CM : r=0.644; data not shown, all p<0.05). NK cells (CD16 + CD56 + ), CD4 + T EM and T EMRA and CD8 + T EM and T EMRA did not correlate with LVEF, and none of these cells express CCR7. We have previously looked at 15 different chemokine receptors expressed on lymphocytes (5), and CCR7 did not correlate with post-reperfusion cell drops. Together, this suggests that the mechanism by which pre-reperfusion leucocyte counts fall in STEMI is specific to CCR7 + cells, and may not occur in CMV seropositive patients.

CD3+ T-Lymphocyte Infiltration in Human Failing Hearts
Finally, we analysed 21 explanted hearts (15 non-failing, 6 failing) for infiltration of CD3 + T-lymphocytes ( Figure 5). In non-failing hearts (8 from CMV seropositive donors, 7 from seronegative), we found a significant increase in CD3 + T-cells in CMV seropositive hearts (2.6 vs 0.75 CD3 + cells/mm 2 , p=0.049), suggesting a role of Tcells for CMV-related myocardial remodeling. Of interest, in failing hearts (all from seronegative donors) the number of infiltrated T-cells appeared even higher (6.4 CD3 + cells/mm 2 , p=0.012 vs non-failing hearts), but there were too few failing hearts from CMV seropositive donors to compare this meaningfully. We did not have access to paired blood samples to measure peripheral blood T-lymphocytes for these patients.
Another limitation is that, due to the small sample of hearts studied, we were not able to adjust this analysis for demographic differences between CMV serogroups. In particular, CMV seropositive patients are generally older ( Table 1), and this may be an important confounding variable in our data.

DISCUSSION
In this study, we describe several observations from varied cohorts. First, we show that remodeling of the left ventricle at 12 weeks after STEMI is worse in CMV seropositive patients.
In a large retrospective cohort, we show that pre-reperfusion lymphopenia is associated with higher troponins and long-term mortality, suggesting that more severe infarcts drive greater lymphopenia even before reperfusion. Detailed analysis of our prospective cohort showed that reduced early LVEF correlated with  reduced counts only of CCR7 + (non-effector) T-lymphocytes, and only in CMV seronegative patients. We propose that CCR7 + Tlymphocytes are depleted prior to reperfusion in CMV seronegative patients, and by a different mechanism CX 3 CR1 + T-lymphocytes are depleted after reperfusion in CMV seropositive patients. Finally, in a small sample of explanted non-failing hearts, we found that lymphocytes were more abundant in hearts from CMV seropositive donorsif replicated in a larger cohort, this would support a hypothesis of heightened myocardial inflammation in these patients.

CX 3 CR1 Is Associated With Post-Reperfusion Lymphocyte Depletion in CMV Seropositive Patients
Having previously shown that post-reperfusion lymphopenia predicts long term mortality in patients following myocardial infarction (5), in this study we show in a very large population of all-comer patients that pre-reperfusion lymphocyte count also predicts mortality, and is inversely correlated with admission troponin. Troponin was used as a universally available surrogate marker of infarct severity, and this association suggests that lymphocytes fall prior to reperfusion, and fall more with larger infarcts and longer ischaemic times (both of which result in high troponin values) (18,19). In our prospective cohort, pre-reperfusion lymphocyte count also correlated with left-ventricular ejection fraction (LVEF), a marker of prognosis and infarct severity, but only in CMV seronegative patients, and only in CD4 + and CD8 + T N and T CM , which express CCR7. Conversely, CD4 + and CD8 + T EM and T EMRA , which are CCR7-negative, were not correlated with LVEF -strongly suggesting that CCR7 + T-lymphocytes are selectively depleted prior to reperfusion. CCR7 is associated with a non-cytotoxic, CX 3 CR1-negative T-cell phenotype and, along with its ligands CCL19 and CCL21, is predominantly involved in trafficking lymphocytes to secondary lymphoid organs (20). Earlier work has shown that CCR7 + T-lymphocytes also drop to some degree in the 30 minutes after reperfusion (5, 21), but we do not know whether these are trafficked to lymphoid organs or sequestered within the myocardial vasculature or myocardium itself. In contrast, T-lymphocyte subsets which are CX 3 CR1 + drop in the 90 minutes after reperfusion. The Bolovan-Fritts group has shown that CMV-specific T-lymphocytes strongly induce membrane-bound fractalkine on vascular endothelium (10), which we hypothesise is interacting with CX 3 CR1+ Tlymphocytes and removing them from circulation. We have previously published evidence that T-lymphocytes are lost within the myocardial vasculature during STEMI/pPCI, and our data from human heart explants now suggests that CMV seropositive hearts display greater infiltration of the myocardium by T-lymphocytes, further supporting the theory that lymphocytes are recruited to the myocardium to a greater extent in seropositive patients (5).

CMV Seropositivity Is Associated With Adverse Left-Ventricular Remodeling After STEMI
The cardiovascular risks conferred by CMV seropositivity remain controversial. Case-control studies, such as that by Siscovick (22), have not identified an association between CMV IgG and the development of cardiovascular disease. As CMV and cardiovascular disease are so prevalent, however, this study design probably lacks power to identify this association, as the authors themselves stated. A 2017 meta-analysis of prospective epidemiological studies estimated a 7-38% increased relative risk of cardiovascular disease in seropositive patients, which the authors calculate would account for 13% of the total cardiovascular disease burden, due to the high prevalence of CMV (23). The impact of CMV on ventricular remodeling after STEMI, however, has not been well studied. We show that CMV seropositive patients display evidence of worse remodeling on cMRI, with ventricular dilatation by 12 weeks after pPCI. This is evidence that latent CMV infection promotes a clinically relevant adverse response to STEMI and pPCI, although the observed effect was small and it is essential that this association is investigated in larger groups of patients, followed over a longer period. A significant association after only 12 weeks could herald a substantial decline in function when measured over years. There is evidence from mouse models of myocardial infarction and reperfusion that a deranged inflammatory response leads to adverse remodeling. The Frangiogiannis group (24) has shown that CCR5 knockout mice are unable to recruit regulatory T-lymphocytes to the infarcted myocardium and display worse remodeling, while Cochain and colleagues showed that the decoy receptor D6, which internalises and defunctions pro-inflammatory chemokines, is necessary for healthy remodeling in the infarcted murine heart (25). We have previously shown that FoxP3 -CD4 + T-lymphocytes significantly contribute to agerelated myocardial inflammation in mice (26). In further studies, we found that CMV-seropositive patients demonstrate signs of accelerated immune ageing following myocardial infarction, that seem to link with impaired myocardial healing (27,28). Accordingly, we suggest that in CMV seropositive human patients, a dysregulated immune response to STEMI and pPCI leads to excessive inflammation, MVO and adverse remodeling.

Anti-Inflammatory Therapy in Myocardial Infarction as a New Therapeutic Target
The growing literature on the inflammatory component of cardiovascular disease reveals an enormous arsenal of potential therapies. To date, however, treatments targeting the immune component of cardiovascular disease have not been universally successful. The CANTOS trial showed for the first time that specific anti-inflammatory therapy (with canakinumab, an interleukin-1b inhibitor) can improve cardiovascular outcomes in patients with a history of myocardial infarction and evidence of persistent inflammation (1), although the reduction in the primary endpoint of major adverse cardiovascular events was modest, driven by lower risk of non-fatal MI, and the lack of a mortality benefit leaves the CANTOS trial as predominantly a proof-of-concept study. Subsequently, the cardiovascular inflammation reduction trial (CIRT) and the colchicine cardiovascular outcomes study (COLCOT) showed that more readily available, less targeted immunosuppressants (low-dose methotrexate and colchicine, respectively), do not meaningfully improve cardiovascular outcomes (2,3). Taken together, these studies support further investigation of well-defined immunological pathways as therapeutic targets in cardiovascular disease, but suggest that non-specific anti-inflammatory therapy is unlikely to succeed.
These trials looked at vascular events, but did not address post-MI outcome. Larger clinical trials on anti-inflammatory approaches in post-MI patients are widely lacking. Gu and colleagues have shown that in a murine model of induced myocardial infarction, specific inhibition of fractalkine improves LV function and survival (29). This specific inhibitor has been trialled in patients with rheumatoid arthritis, and was safe and well-tolerated (30). Specific blockade of the fractalkine-CX 3 CR1 axis in humans therefore emerges as a tantalising candidate for a novel adjunctive therapy in STEMI patients treated with pPCI, which may improve left ventricular function and reduce microvascular obstruction, particularly if targeted to CMV seropositive patients.

Conclusion
Our study provides evidence that the healing process in myocardial infarction is different for patients who are seropositive for previous cytomegalovirus infection. We suggest that additional myocardial inflammation triggered by cytotoxic T-lymphocytes and CX 3 CR1 is more detrimental to recovery, making fractalkine signalling a valid drug target in seropositive patients.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by National Research Ethics Committee North-East -Newcastle and North Tyneside 2; REC reference: 14

AUTHOR CONTRIBUTIONS
LS performed analysis of cellular and cMRI data, and wrote the manuscript, with revisions from GR, UH, KSt and IS. cMRI scans were performed and analysed by CP, AM and SB. Cellular data was collected and analysed by CP, SC, PP, SB, KB, KSo and GR. Analysis of explanted human hearts was performed by VS, LR and PR. The study was led by IS. All authors contributed to the article and approved the submitted version.