K. Ozawa, Hepatic Disease Research Institute, Honmachi-15chyome-754-1, Higashiyama-ku, Kyoto 605-0981, Japan. E-mail: email@example.com
This study investigated how CD8+ T cell subsets respond to allo- and infectious immunity after living donor liver transplantation (LDLT). Early alloimmunity: 56 recipients were classified into three types according to the post-transplant course; type I demonstrated uneventful post-transplant course, type II developed severe sepsis leading to multiple organ dysfunction syndrome or retransplantation and type III with acute rejection. In 23 type I recipients, the interleukin (IL)-12 receptor beta-1 (Rβ1)+ cells of central memory T cells (Il-12Rβ1+ TCM) were increased above the pretransplant level. In 16 type II recipients, IL-12Rβ1+ TCM was decreased markedly below the pretransplant level on postoperative day (POD) 5. In 17 type III recipients, IL-12Rβ1+ TCM was decreased for a more prolonged period until POD 10. Along with down-regulation of IL-12Rβ1+ TCM, the IL-12Rβ1+ cells of CCR7-negative subsets (CNS) as well as perforin, interferon (IFN)-γ and tumour necrosis factor (TNF)-α decreased gradually, resulting in the down-regulation of effectors and cytotoxicity. The down-regulation of IL-12Rβ1+ TCM was suggested to be due to the recruitment of alloantigen-primed T cells into the graft, and then their entry into the secondary lymphoid organ, resulting in graft destruction. Infectious immunity: immunocompetent memory T cells with the capacity to enhance effectors and cytotoxicity were generated in response to post-transplant infection along with both up-regulation of the IL-12Rβ1+ TCM and an increase in the CNS showing the highest level of IL-12Rβ1+ cells. In conclusion, this work demonstrated that the IL-12Rβ1+ cells of TCM and CNS are regulated in a tightly coupled manner and that expression levels of IL-12Rβ1+ TCM play a crucial role in controlling allo- and infectious immunity.
In transplant settings, a critical event during the progressive destruction of graft tissue is the recruitment of alloantigen-primed T cells into the allograft through stimulation of allogeneic endothelial cells and their migration to secondary lymphoid organs . The facets of CCR7 subsets depend upon T cells invested with distinct homing and effector capacities [2,3]. The contribution of CCR7(+) memory cells expressing lymph node-homing receptors – central memory T cells (TCM) and CCR7(−) memory cells expressing receptors for migration to inflamed tissues – effector memory T cells (TEM) to allograft rejection or destruction has been suggested. Recently, it has been emphasized that CD8+ TCM or TEM is the principal memory subset responsible for allograft rejection.
Immediately after living donor liver transplantation (LDLT), activated dendritic cells (DCs) produce a variety of cytokines such as interferon (IFN)-γ, tumour necrosis factor (TNF)-α and interleukin (IL)-12, which can provide an important regulatory signal for naive CD8+ T cells (TN) for activation and proliferation as well as for differentiation into CD8+ effector T cells (TE). Based upon considerable accumulated data, it has been considered that IL-12 produced primarily by DCs plays the most important role in promoting T helper type 1 (Th1)-type immune response and cell-mediated immunity . IL-12 receptor beta-1+ cells (IL-12Rβ1+ cells) are essential for IL-12-driven enhancement of alloimmune responses [7–9].
Our previous work showed that transplant recipients with the highest numbers of pre-existing TE developed a high incidence of infection and showed a poor prognosis for survival after LDLT compared with recipients exhibiting low pre-existing TE levels [10,11]. The pre-existing highest TE seems to be determined by a host history of previous infections, and potentially cross-react with allogeneic MHC molecules, providing a potent barrier to tolerance and determining the outcome of infection (so-called heterologous immunity) [12,13]. Accordingly, based on this sequence of events, this study clarifies further the fundamental mechanism underlying changes in phenotypic and functional features in CD8+ T cell subsets after LDLT in recipients with the lowest pre-existing TE level.
Furthermore, in order to clarify the features of all CD8+ T cell subsets, double-positive (CD45RORA) CCR7(+) T cells (CD45RORACCR7+) (TDP+) and double-positive CCR7(−) T cells (CD45RORACCR7-) (TDP−) were analysed, in addition to the four major subsets of TN, TCM, TEM and TE. Here, we investigated how CD8+ T cell subsets are generated functionally along with the expression levels of IL-12Rβ1+ cells of CD8+ TCM (IL-12Rβ1+ TCM) and CCR7-negative subsets (CNS) after LDLT.
Patients and methods
We examined 56 recipients who had undergone our standard LDLT  between 2004 and 2009 at Kyoto University Hospital and had exhibited the lowest pre-existing TE levels, so-called Group I recipients, classified previously by hierarchical clustering . Written informed consent was obtained from the recipients before starting the study, which was approved by the Ethics Committee of Kyoto University Hospital and conducted in accordance with the 1975 Declaration of Helsinki, as revised in 1996.
Methylprednisolone (initial steroid bolus; 10 mg/kg) was administered just prior to the start of graft reperfusion, as shown previously .
Regular immunosuppression protocol using tacrolimus (Tac) and corticosteroids was performed routinely from postoperative day (POD) 2 . For ABO-incompatible LDLT, protocol using rituximab prophylaxis and prostaglandin E1 through the hepatic artery and systemic cyclophosphamide, followed by mycophenolate mofetil (MMF), was performed in addition to the standard protocol .
Definition and treatment of acute graft rejection
Percutaneous liver biopsy was performed in the event of clinical or laboratory signs of acute graft rejection and specimens were graded according to the Banff criteria .
Definition of an infectious complication
A bacterial, viral or fungal infection was assumed to have developed if clinical and/or laboratory evidence consistent with acute infection were detected. Such laboratory evidence included relevant positive serological markers and cultures . The criteria for sepsis defined by Bone were applied .
Serological tissue typing for human leucocyte antigens (HLA) A, B (Bw), C, DR and DQ for classes I and II loci was undertaken in all patients.
We examined peripheral blood mononuclear cells (PBMCs) from each recipient. Sample analyses were performed within 24 h after sampling in all cases. Because the numbers of CD8+ T cells often decreased postoperatively to fewer than 10% lymphocytes, we had to analyse CD8+ T cell subsets with low numbers of events. Consequently, we always performed at least duplicate assays of the same sample. Cell staining was undertaken using monoclonal antibodies as reported previously . Monoclonal antibodies used to stain cell surface antigens were as follows: allophycocyanin (APC; Coulter Immunotech, Miami, FL, USA) or PC-5 (Coulter Immunotech, Marseilles, France)-conjugated anti-CD4 or CD8, fluorescein isothiocyanate (FITC)-conjugated anti-CD45RO (Nichirei, Tokyo, Japan), TC-conjugated anti-CD45RA (Caltag Laboratories, Burlingame, CA, USA), phycoerythrin (PE)-conjugated anti-CD3 (Coulter Immunotech, Miami), FITC-conjugated anti-CD19 (Coulter Immunotech, Marseilles) and PE-conjugated anti-human CCR7 (Dako Cytomation, Kyoto, Japan).
Flow cytometric detection of cytokine production and intracellular staining for perforin
Flow cytometric measurement of IFN-γ and TNF-α production was performed as described previously . IL-2 production was measured using FITC-conjugated anti-Hu-IL-2 (BD Bioscience, San Jose, CA, USA). Cells were stimulated with a mixture of phorbol myristate acetate (PMA) (25 ng/ml; Sigma-Aldrich Chemical Co., St Louis, MO, USA) and ionomycin (1 µg/ml; Sigma-Aldrich) with the Golgi inhibitor brefeldin A (10 µg/ml; Sigma-Aldrich). We measured intracellular perforin in CD8+ T cells without previous stimulation. Perforin analysis was performed according to the previously reported method [10,19].
Expression of IL-12 receptors was determined using R-PE-conjugated anti-IL-12Rβ1 and IL-12Rβ2 (BD Biosciences, San Diego, CA, USA). The IL-12Rβ1+ cells in six CD8+ T cell subsets were measured after classification of CD8+ T cells into three subsets, as shown in Fig. 1.
Evaluation of post-transplant immune status
Post-transplant changes in phenotypic and functional properties of CD4+ and CD8+ T cells, albeit at different periods, could be almost restored to pretransplant patterns. As a measure, the proportions of the variables immediately before LDLT were subtracted from the proportions at various time-points after LDLT, and expressed as % differences . Similarly, the % differences were calculated for other variables such as IFN-γ, TNF-α, IL-12Rβ1+ cells and perforin.
To select the recipients with the lowest pre-existing TE (so-called Group I), we performed a hierarchical cluster analysis  using jmp 7 (SAS Institute Inc., Cary, NC, USA), as reported previously .
Comparisons for continuous variables between groups were performed by applying Student's t-test and analysis of variance. Comparisons for proportions between groups were undertaken using Fisher's exact test or χ2 test. All statistical tests were two-tailed. Significance was defined as P < 0·05.
Clinical analyses and classification of three types according to post-transplant episodes
Table 1 shows the clinical analyses of 56 recipients, who were classified into three types based on post-transplant episodes: 23 type I recipients demonstrating uneventful courses during the post-transplant period, although bacterial or cytomegalovirus (CMV) infections were often encountered; 16 type II recipients demonstrating severe sepsis leading to multiple organ dysfunction syndrome (MODS) or retransplantation; and 17 type III recipients demonstrating acute rejection (15 cellular and 2 humoral). Patient ages were slightly higher in type I than in types II and III. Among primary diseases, the incidence of viral hepatitis C (HCV) infection was highest in type I. There was no difference in Model for End-stage Liver Disease (MELD) score , HLA-mismatch numbers and numbers of ABO-incompatible LDLT among the three types. Among the maintenance immunosuppressive regimens, the incidence of combined Tac plus corticosteroid (T/C) and Tac plus MMF plus corticosteroid (T/M/C) was not different among three types. Interestingly, the period of hospital stay until discharge was shortest in type I, longest in type II and intermediate in type III groups.
Table 1. Clinical analyses of three types classified according to post-transplant episodes.
Immunosuppressants: T, tacrolimus; M, mycophenolate mofetil (MMF); C, corticosteroid; CsA, cyclosporine A; SS/MODS, severe sepsis combined multiple organ dysfunction syndrome; Re-Tx, re-transplantation. *Mean ± standard deviation. HLA: human leucocyte antigen; IFN: interferon; IL: interleukin; MELD: Model for End-Stage Liver Disease; TNF: tumour necrosis factor.
50·72 ± 9·23
8·45 ± 4·30
17·76 ± 7·18
7·51 ± 5·82
5·60 ± 2·94
10·17 ± 5·20
48·26 ± 10·87
46·26 ± 12·96
57·12 ± 10·32
61·52 ± 21·76
5·54 ± 3·66
16·17 ± 13·66
4·04 ± 3·36
5·99 ± 4·58
6·94 ± 4·66
31·84 ± 21·55
35·70 ± 16·45
41·33 ± 18·80
62·24 ± 13·19
6·93 ± 3·57
14·96 ± 9·64
3·41 ± 2·51
6·35 ± 2·70
6·31 ± 3·25
33·99 ± 14·69
37·86 ± 12·95
37·98 ± 13·82
In the pre-existing levels of CD8+ T cell subsets, the proportion of TN was slightly, but significantly, decreased compared with that of types II and III. The proportion of TEM, TDP− or TNF-α was significantly high in type I. The pre-existing levels of TCM, TE, TDP+ and TDP− as well as IL-12Rβ1+cells and IFN-γ in CD8+ T cells were not different among the three types.
Post-transplant changes in the percentage difference of CCR7-negative (CNS) and -positive subsets (CPS) related to IL-12Rβ1+ TCM, and IL-2 production of CD4+ TCM and TEM after Tac administration
Figure 2a shows changes in the % difference in both CNS and CPS related to the levels of IL-12Rβ1+ TCM after LDLT in each of the three types. In type I recipients, IL-12Rβ1+ TCM remained at the pretransplant level until POD 5 and then increased in response to infection. Post-transplant changes in CNS and CPS each showed a small range around zero until POD 20 and then changed markedly. These recipients were uneventful during the post-transplant period, although they often developed slight infections. In type II recipients, IL-12Rβ1+ TCM was decreased to approximately −30% on POD 5 and then returned to around pretransplant level on POD 7. CNS was slightly increased despite changes in IL-12Rβ1+ TCM. In type III recipients, IL-12Rβ1+ TCM was decreased markedly for a prolonged period (from POD 2–10) and then increased to the pretransplant level after POD 10. CNS remained at the pretransplant level until POD 5 and then returned to a higher level (approximately 20%). In all three types, changes in the CNS were correlated inversely with those in CPS.
Figure 2b shows changes in IL-2 expression in CD4+ TCM and TEM after LDLT and Tac administration in 15 recipients. The % difference in IL-2 expression of CD4+ TCM, although slight in TEM, was suppressed continuously approximately −25% from POD 5, along with an adjustment of Tac trough levels to an appropriate level. These findings suggest that in types II and III, IL-12Rβ1+ TCM was down-regulated despite complete inhibition of IL-2 production by Tac.
Changes in CD8+ T cell subsets and their IL-12Rβ1+ cells after LDLT in a representative type I recipient
Figure 3a shows changes in six subsets of CD8+ T cells after LDLT in a representative type I recipient (a 50-year-old woman) undergoing ABO-incompatible LDLT for primary biliary cirrhosis. High white blood cell counts greater than 20 000/mm3 persisted between PODs 4 and 12, along with increased levels of C-reactive protein (5–10 mg/l). CMV infection occurred on POD 29. Bacteria were detected in bile on PODs 30 and 33. The total bilirubin level increased to greater than 10 mg/dl from PODs 12–16 with a rise in aspartate aminotransferase (>100 IU/l).
The proportion of TE increased gradually from 15% at time 0 to 38·5% on POD 33; significant inverse changes in TN were also detected (r = −0·975, P < 0·005) (Fig. 3i and ii). Other subsets remained unchanged near pretransplant proportions.
Figure 3b shows changes on flow cytometry in the IL-12Rβ1+ cells in each of the CD8+ T cell subsets after LDLT. The IL-12Rβ1+ cells are expressed as % proportion of IL-12Rβ1+ cells included in each subset. IL-12Rβ1+ cells were expressed in a markedly higher percentage in CD8+ CNS compared with those of the CPS (Fig. 3vi). Those in CNS were decreased slightly on POD 5 but returned promptly to pretransplant levels before POD 12. Among the CPS, IL-12Rβ1+ TCM were greatly increased along, with the severity of infectious complications. The changes in IL-12Rβ1+ cells of the CNS and CPS were depicted clearly by the % difference (Fig. 3vii and viii). Importantly, TE generation was correlated highly (r = 0·829, P = 0·083, n = 5) and positively with the increase in IL-12Rβ1+ TCM (Fig. 3ii and viii).
Changes in CD8+ T cell subsets and their IL-12Rβ1+ cells after LDLT in a representative type II recipient
Figure 4a shows changes in the % proportion of six subsets of CD8+ T cells on flow cytometry after LDLT in a representative type II recipient (a 53-year-old male) undergoing ABO-incompatible LDLT due to both HCV and HBV-related liver cirrhosis.
He was complicated by microbial infection and sepsis associated with elevated body temperature (>39°C), high white blood cell counts (>15 000/mm3) and high C-reactive protein (>10 mg/l) between POD 21 and 25 and again between POD 65 and 91. Afterwards, MODS developed along with severe pneumonia (requiring reintubation) on POD 25 and then severe sepsis (SS)/MODS developed with life-threatening infectious complication on POD 87. Thereafter, the patient died on POD 100. Microbial infection persisted for a prolonged period from POD 6 (B6) to the terminal stage (B89). CMV infection was detected on PODs 12, 40 and 50. Three immunosuppressants consisting of Tac, MMF and rapamycin had been administered after LDLT. Tac was administered at a low trough level of 5 (ng/ml) from POD 49 and was then stopped on POD 66, because of severe infectious complications. Rapamysin was administered instead. MMF was administered between POD 63 and 97.
The six CD8+ T cell subsets were unchanged till POD 19 and then TDP− increased to the highest level with TE at the development of infection and sepsis (i). However, TEM increased transiently between PODs 26 and 42 and then decreased to the lowest level. In contrast, in the CPS, TN was decreased markedly from POD 19, whereas TDP+ and TCM remained unchanged at the lowest level during the post-transplant period. Importantly, TDP− and TE increased markedly as the severity of infection progressed from simple microbial infection through sepsis to SS/MODS. The CNS remained at the highest level even during the terminal stage prior to death. Those events were more depicted clearly by the % difference (ii).
Figure 4b shows changes on flow cytometry in IL-12Rβ1+cells in each of the CD8+ T cell subsets after LDLT. In the % proportion (Fig. 4vi), the IL-12Rβ1+ TCM decreased markedly from 60% at 3 h to 20% at 6 h and then persisted at the same level till POD 5. Thereafter, those values were restored at least to pretransplant level until POD 12 after Tac administration, followed by further increase. That of TDP+ also decreased slightly to the lowest level between 6 h and POD 5, and then increased above the pretransplant level from POD 26. However, IL-12Rβ1+ cells in the CNS gradually decreased after 6 h and reached the lowest levels on POD 5 at approximately 25% in TE and TDP− and 50% in TEM. Thereafter, the IL-12Rβ1+ cells in TE, TEM and TDP− were up-regulated promptly to 100% on POD 12 and then remained at the highest level during the post-transplant period. These events were depicted more clearly using the % difference (Fig. 4vii and viii). The % difference of IL-12Rβ1+ TCM (% IL-12Rβ1+ TCM difference) was increased from POD 19, with the development of sepsis (Fig. 4viii). More importantly, the decrease in IL-12Rβ1+ TCM preceded the down-regulation of those in the CNS, but both were almost restored to pretransplant level by POD 12.
Changes in the expression of perforin, IFN-γ and TNF-α of the CNS in relation to IL-12Rβ1+ TCM
Figure 5 shows changes on flow cytometry in the % difference of perforin, IFN-γ and TNF-α in TE, TEM and TDP− compared with IL-12Rβ1+ TCM after LDLT in the same recipient as shown in Fig. 4. In three subsets, the expression of perforin, IFN-γ and TNF-α was down-regulated significantly similarly by POD 2, except for TNF-α of TEM, along with a decrease in IL-12Rβ1+ TCM. Thereafter, the three variables tended to be up-regulated from POD 12, with an increase in IL-12Rβ1+ TCM except for TNF-α in TEM.
Changes in CD8+ T cell subsets and their IL-12Rβ1+ cells after LDLT in a representative type III recipient
Figure 6 shows changes in the % proportion of six subsets of CD8+ T cells on flow cytometry after LDLT in a representative type III recipient (a 36-year-old female) undergoing ABO-compatible LDLT due to fulminant hepatic failure. She was immunosuppressed by the combined therapy of Tac and corticosteroid. She had a high risk (donor+/recipient-) of CMV. On POD 11, the preservation injury was conformed by histological examination of the biopsy specimen and then the biopsy-proven mild ACR was detected on POD 23. CMV infections were detected on PODs 13, 17 and 20. Bacterial infection by P. aeruginosa was found in catheter on POD 4.
The CNS remained unchanged at the lowest level until POD 12. On POD 28, TE was increased to 40%, TDP− to 20% and TEM remained unchanged at the lowest level (Fig. 6i). In IL-12Rβ1+ cells of CD8+ T cell subsets, those of CNS were down-regulated on PODs 3–5, to the lowest in TDP−, slightly in TEM and to pretransplant level in TE. The IL-12Rβ1+ cells of CNS were down-regulated along with IL-12Rβ1+ TCM and then returned to the highest level with their restoration. The IL-12Rβ1+ TCM was decreased markedly to −25% on POD 3 and then increased gradually to the pretransplant level on POD 28, due possibly to preservation injury and then ACR (Fig. 6vi and viii). In contrast, IL-12Rβ1+ TDP+ was increased to approximately 10% above pretransplant level from POD 7. Importantly, TE generation was increased remarkably on POD 28, when IL-12Rβ1+ TCM was up-regulated around pretransplant level.
Infiltration of alloantigen-activated CNS into graft endothelium and the down-regulation of IL-12Rβ1+ TCM
As soon as the largest numbers of donor-specific alloantigens were released from the allograft immediately after LDLT or during acute rejection, the CD8+ T cells of the recipient are primed by the alloantigen, and encounter allogeneic endothelial cells during infiltration in the graft. The activated donor-specific memory T cells lose CCR7 function and directly mediate lytic destruction in the graft, and consequently enter secondary lymphoid organs by entering afferent lymphatic vessels under the tightly regulated expression of CCR7 [22,23]. During this process, the CNS produced down-regulates the CCR7 function transiently by responding actively to TCM circulating exclusively through secondary lymphoid organs  and induces the potential to produce effector and cytokines during their sequestration, due possibly to the up-regulation of IL-12Rβ1+ cells [7–9]. It can be suggested that the loss of CCR7 and the acquisition of effector function are regulated co-ordinately and tightly by the level of IL-12Rβ1+ cells in some cells residing in the TCM. Consequently, this may facilitate the accumulation of alloantigen-experienced CNS at the allograft while also allowing the CNS to exit rapidly and continue their immune surveillance elsewhere.
Specific characteristics of types II and III recipients
In type I, the IL-12Rβ1+ TCM remained unchanged immediately after LDLT and were then up-regulated over the pretransplant level in response to post-transplant infection. However, the characteristics of types II and III are shown as follows.
First, in type II, the IL-12Rβ1+ TCM was down-regulated rapidly to the lowest level below pretransplant level on POD 5 and continued for 7 days, whereas in type III, IL-12Rβ1+ TCM was down-regulated more progressively for a more prolonged period. This marked down-regulation seems to be due to excessive accumulation of alloantigen-activated CNS in the secondary lymphoid organ through the graft (Fig. 2). Recently, we found that IL-12Rβ1+ TCM is down-regulated below pretransplant level during progression of acute rejection despite administration of the presence of immunosuppressive maintenance regimens (to be submitted).
Secondly, the IL-12Rβ1+ cells of all circulating CNS were maintained at the highest levels in type I. In type II, in contrast, the IL-12Rβ1+ cells of the CNS were down-regulated gradually after a rapid decrease in IL-12Rβ1+ TCM. Initially, alloreactive CNS were equipped with the highest IL-12Rβ1+ cells, facilitating their ability to stage an efficient immune response promptly by enhancing effectors and cytotoxicity in the allograft. Thereafter, the IL-12Rβ1+ cells of the CNS gradually decreased markedly along with decreases in perforin, IFN-γ and TNF-α, resulting in the functional impairment of CD8+ T cells. It is likely that decreases in IL-12Rβ1+ cells in TCM and CNS occurred when alloantigen-specific cells remain sequestered within secondary lymphoid organs, before their re-entry into the recirculating lymphocyte pool. This period demonstrates the highest susceptibility to life-threatening infectious complications.
Thirdly, in addition to the preceding down-regulation of IL-12Rβ1+ TCM, the down-regulation of IL-12Rβ1+ cells of the CNS was restored completely immediately after up-regulation of IL-12Rβ1+ TCM. The IL-12Rβ1+ TCM may regulate that of the CNS within the secondary lymphoid organs.
Lastly, Tac inhibits the production of IL-2, thus leading to a decrease in the proliferation of activated lymphocytes . However, the down-regulation of IL-12Rβ1+ cells of TCM and CNS continued despite complete inhibition of IL-2 by Tac. These results are in line with previous reports, that there is no reliable approach to preventing the deleterious effect of alloreactive memory T cells in transplant settings .
Development of immunocompetent alloreactive memory T cells along with an increase in IL-12Rβ1+ TCM
IL-2-independent generation of the CNS occurred along with an increase in IL-12Rβ1+ TCM above pretransplant level in all three types. Even in type II, both effectors and cytotoxicity of CD8+ T cells increased as the severity of infection progressed from microbial infection through sepsis to SS/MODS, along with increases in IL-12Rβ1+ TCM. Thus, the increase in IL-12Rβ1+ TCM may induce fully functional CNS populations, with the highest IL-12Rβ1+ cells capable of immediate synthesis of perforin when a newly invading antigen is encountered despite inhibition of IL-2 production. These sequential events may be related closely to the self-renewing cells residing in TCM, as shown by Fearon et al. .
Finally, these results suggest that the recipients with a marked down-regulation of IL-12Rβ1+ TCM showed the highest incidence of severe post-transplant complication, although IL-12Rβ1+ TCM was up-regulated in response to post-transplant infection. It seems likely that the level of IL-12Rβ1+ TCM plays an important role in determining the clinical outcomes in the inhibition of IL-2 production by Tac.
This work was funded in part by grants (numbers 13204041 and 13307038) from the Scientific Research Fund of the Ministry of Education, Science and Culture of Japan. We thank members of the Department of Surgery of Kyoto University Medical School for their assistance in this work.
None of the authors has any conflict of interest with the subject matter or materials discussed in the manuscript.