1. Introduction
Cervical cancer is the second most prevalent tumor in developing countries and the fourth most common cause of cancer-related deaths among women. Over half a million new cervical cancer cases, and an estimated 265,700 deaths are reported each year worldwide [
1,
2]. Several factors including economic conditions, genetic factors, endocrine [
3], and immunity play significant role in the progression of cervical cancer. High-risk human papilloma virus is the primary cause of cervical cancer (CC) [
4]. Immunosuppression states like infection with HIV [
5] or taking immunosuppressive drugs [
6] increases susceptibility to HPV infection and which subsequently causes cervical cancer. The current cervical cancer treatments include surgery, chemotherapy, and radiotherapy, but these are not effective for the management of advanced local cervical cancer, metastatic and recurrent tumors [
7]. Recently, immunotherapy, particularly the immune checkpoint inhibitors, has achieved a great breakthrough. For example, the use of Pembrolizumab, a PD-1 inhibitor, in the later-staged and recurrent CC was authorized [
8]. However, not all CC patients are sensitive to Pembrolizumab. Additional immune-related molecules should, therefore, be identified to explain the pathogenesis and improve the treatment of cervical cancer.
The role and mechanism of immune cells in the development of cervical cancer has not been adequately studied. Previous studies showed that the number of FOXP3+ regulatory T-cells that could suppress the innate and adaptive immunity systems [
9,
10] was higher in cervical cancer compared to other types of tumors [
11] and suppressed immune responses [
12]. On the one hand, it was reported that cervical cancer cells could secrete indoleamine 2,3-dioxygenase (IDO) to recruit FOXP3+ regulatory T-cells [
12]. On the other hand, cervical cancer cells can secrete a series of molecules such as PEG2, IL-6, CCL2, and IL-10 that differentiate and activate M2-like macrophages [
13,
14,
15,
16]. M2-like macrophages have been widely accepted to play a role in the poor prognostic effect in CC [
17]. M2-like macrophages promoted the CC cell proliferation by the GM-CSF/HB-EGF paracrine loop [
18]. The number of M2-like macrophages was related to invasion patterns [
19] and lymph node metastasis [
20]. M2-like macrophages participate in immune suppression in cancer [
21], yet, its underlying mechanism in CC has not been sufficiently elucidated. Previous studies on cervical cancer indicated that M2-like macrophages could decrease the presence of HPV16 E7 specific CD8+ T cells by diminishing HLA-DR expression and increasing the expression of either IL-10 or CSF1R [
22,
23,
24]. Activated macrophages may inhibit the number of CD4+ T cells by producing neopterin [
25]. M2-like macrophages could decrease the percentage of HPV specific regulatory T cells by blocking IL-10 signaling [
26]. However, the association of CCL22 from cervical cancer cells or macrophages and regulatory T cells is still elusive.
The C-C motif chemokine ligand 22 (
CCL22) gene is a secreted protein that exerts chemotactic activity for monocytes, dendritic cells, natural killer cells, and for chronically activated T lymphocytes [
27,
28,
29,
30]. Accumulating studies indicate that CCL22 plays a tumor-promoting role in human cancer. In ovarian cancer, for instance, CCL22 was found to induce regulatory T (T-reg) cells into tumor mass and inhibit T cell immunity [
31]. High expression of CCL22 in M2 macrophages confers resistance to 5-fluorouracil in colorectal cancer [
32]. A previous study showed that the
CCL22 mRNA expression level was higher in CC tissue than in a normal cervix [
33]. However, the function of CCL22 in cervical cancer remains unknown.
The present study determined the functional role of CCL22 in infiltrating macrophages in cervical cancer. The expression level of CCL22 and the FOXP3+ regulatory T-cell marker was measured using a tissue microarray (TMA) with immunohistochemical staining. We further evaluated the correlation between clinical characteristics and CCL22 and FOXP3 expression. The findings of our study indicated that the number of CCL22+ cells was positively correlated with that of FOXP3+ cells (r = 0.210, p = 0.001). Moreover, group CCL22high had a significantly lower overall survival rate (OS), compared to the CCL22low group (p = 0.001). There was, however, no significant difference in progression free survival (PFS). The OS of the combination group CCL22highFOXP3high was significantly lower than that of group CCL22lowFOXP3low regardless of the FIGO stage and disease subtype (p < 0.05). The PFS of group CCL22highFOXP3high was significantly lower than that of group CCL22lowFOXP3low in cervical adenocarcinoma (p < 0.05). A double immunofluorescence staining indicated that M2-like macrophages primarily secreted CCL22. These results suggest that CCL22 secreted by M2 macrophages could recruit T-reg cells in cervical cancer and reduce the patient survival rate.
3. Materials and Methods
3.1. Bioinformatics
The GEPIA database was used to identify the expression profile of CCL22 (
http://gepia.cancer-pku.cn/) [
34]. Transcript expression analysis for CCL22 was carried out in a total of 316 samples including 13 normal and 303 CESC tissues across TCGA normal and GTEx data. Next, using ANOVA method, 1 as the Log2FC cutoff value, 0.01 as the cutoff value of the significance level, the expressing difference of CCL22 in CESC tissue was obtained. TIMER database was used to identify the correlation between T-regs, TAM2, and CCL22 (
https://cistrome.shinyapps.io/timer/) [
35]. UALCAN database was used to analyze survival rate in groups with differently expressed CCL22 in CESC dataset from TCGA (
http://ualcan.path.uab.edu/index.html) [
37].
3.2. Clinical Sample
The Institutional Review Board of the Ludwig-Maximilian University, Munich, (Number of approval: 337-06, 29 December 2006) approved this study. All patients provided written informed consent. Human cervical cancer tissue microarrays contained 230 samples from the patients diagnosed with CC between January 1994 and September 2002 consecutively from the Department of Obstetrics and Gynecology, LMU Munich. The age was from 22 to 83 years with a median age of 49 years. Clinicopathologic characteristics of patients are available in
Table 1. Staging was performed according to the FIGO staging system. Primary treatment of CC consisted of type 3 radical hysterectomy with pelvic lymph-node dissection. Data including age, stage, lymph node metastasis, distant metastasis, and survival status were obtained from medical records. Data on tumor size, tumor grade, and lymph-node metastases were reviewed from pathology reports.
3.3. Tissue Microarray Construction
TMAs were constructed from 230 formalin-fixed, paraffin-embedded cervical cancer tissue specimens, including 187 cervical squamous cancer and 43 cervical adenocarcinoma. Briefly, to define representative tumor areas, an institutional pathologist reviewed the hematoxylin and eosin (HE)-stained whole-mount sections to define representative tumor areas. Subsequently, after retrieval of four 1.0-mm diameter tissue cores from the formalin-fixed paraffin-embedded tissue blocks of which consisted tumor specimens, they were arrayed on a 38 × 25 recipient paraffin block using an MTA-1 manual tissue array (Beecher Instruments Inc., Silver Spring, MD, USA). Then, 3 μm sections were cut on a microtome and placed on glass slides for two replicates. One series was for staining CCL22 and the other was for staining FoxP3. The presence of tumor tissue on the sections was verified by HE staining.
3.4. Immunohistochemistry
The TMA sections were deparaffinized in xylol for 20 min and rinsed in 100% ethanol. Then, 20 min H
2O
2 incubation was performed to inhibit endogenous peroxidase reaction. Afterwards, the specimens were rehydrated in deescalating alcohol gradients, starting with 100% ethanol and ending with distilled water. The pressure pot contained a sodium citrate buffer (pH = 6.0), which consisted of 0.1 m citric acid and 0.1 mM sodium citrate in distilled water. Then, samples were washed in PBS twice and incubated with a blocking solution (Regent 1, ZytoChem Plus HRP Polymer System (Mouse/Rabbit), Zytomed, Berlin, Germany) for 5 min. Incubation with the primary antibody was performed at 4 °C for 16 h. All antibodies used are listed in
Table 4. Samples were washed twice in PBS (pH = 7.4) following every subsequent step. The blocking solutions (Regent 2) was applied for 20 min and HRP-Polymer (Regent 3) for 30 min. The chromogen-substrate staining was carried out using the Liquid DAB+ Substrate Chromogen System (Dako Scientifi, Glostrup, Denmark). The reaction was stopped by applying distilled water. Finally, the slides were counterstained with Hemalaun for 2 min and blued in tap water. The slides were dehydrated in an ascending alcohol gradient and cover slipped with Eukitt quick hardening mounting medium (Sigma Aldrich, St.Lousis, MO, USA). Human tonsil tissue was applied as positive control. All slides were analyzed using the microscope Leitz Wetzlar (Wetzlar, Germany; Type 307-148.001 514686). The immunoreactive score (IRS) was used for evaluation of the intensity and distribution pattern of antigen expression. The semiquantitative score was calculated as follows: the optical staining intensity (grades: 0 = none, 1 = weak, 2 = moderate, 3 = strong staining) was multiplied by the total percentage of positively stained cells (0 = none, 1 = 10%, 2 = 11–50%, 3 = 51–80%, and 4 = 81% of the cells). This multiplication has a minimum of 0 and a maximum of 12. Two experience staff members analyzed the slides independently. Total number of CCL22+ cells and FOXP3+ cells were counted in a magnification field of 40× lens three times each in three different areas of each slide. The cutoff value was calculated by ROC curve. Cutoff values of 11 and 27 were applied for CCL22+ cells and FOXP3+ cells, respectively.
3.4.1. Evaluation of CCL22+ Cells as Macrophages
For the visualization of CCL22 expressing cells in CC infiltrating cells, the same tissue samples with TMA were used. The antibodies are shown in
Table 4. Double immunofluorescence staining for CCL22 and CD68 as a specific macrophage marker was performed to identify the CCL22 expression in macrophages.
3.4.2. Evaluation of CCL22+ Cells as M2-Like Macrophages
In order to identify which subtype of macrophages express CCL22, the same tissue samples with TMA were used. The antibodies are shown in
Table 9. Double immunofluorescence staining for CCL22 and CD163 as a specific M2-like macrophage marker was performed.
3.5. Cell Coculture
To determine the effect of cervical cancer cells on the induction of CCL22 in macrophages, monocytes were cocultured with HeLa cells and Caski cells, respectively. Transwell core size was 0.4 μm. First, we picked up the ordered buffy coat and purified the PBMCs using the Ficoll–Hypaque method. We counted the PBMCs and seeded 0.2 million PBMCs per well into 24-well plates. Then plastic adherence was in the incubator overnight. The next day, adhered monocytes were washed gently with 1 × PBS, digested by 0.25% trypsin with 0.25% EDTA and ended by complete medium. Monocytes were plated in the lower chamber and 107 cervical cells were plated in the upper chamber in 24 wells. They were cultured in 500 μL culture medium in each chamber with RPMI 1640 medium and 10% fetal calf serum. Cocultures were incubated for 48 h and the supernatant of the lower chambers were harvested followed by Elisa. 0.2 million monocytes with inserts and 107 Caski and 107 HeLa cells with inserts cultured with RPMI 1640 medium with 10% fetal cow serum were as the control. Elisa was performed after 48 h.
3.6. Elisa
Serum of all groups were collected and stored in −80 °C before analysis. ELISA assay was performed according to the manufacturer’s instructions of the ELISA kits (Catalog No. DMD00, R&D Systems, Minneapolis, MN, USA). The OD value at 450 nm was measured. The concentrations of CCL22 was calculated according to the standard curve.
3.7. Statistical Analysis
Clinical data are presented as number (percentage of the amount of relevant clinicopathology). Statistical analysis of the number of CCL22+ and FOXP3+ cells were performed using chi-square test. Correlation analysis of the number of CCL22+ and FOXP3+ cells were performed using Spearman correlation. The Kaplan–Meier method was used to calculate the OS curve and further survival analysis was performed using the log-rank test. The Cox proportional hazard model was used to estimate hazard ratios and 95% confidence intervals in both univariate and multivariate models. CCL22 concentration was presented as mean ± standard deviation. ANOVA analysis was performed using Welch ANOVA and Games-Howell. Statistical analyses were performed using SPSS version 23.0 (SPSS Inc., Chicago, IL, USA). p < 0.05 was considered statically significant.
4. Discussion
Our study revealed a correlation between the expression of CCL22 and FOXP3 in cervical cancer. This study is the first to report the relationship between CCL22 expression and the prognosis of cervical cancer (CC) patients. The group with a higher number of CCL22+ infiltrating cells shows a lower overall survival rate (OS) than that of the group of lower number of CCL22+ infiltrating cells in cervical cancer. However, the expression of CCL22 in cervical cancer cells seems not to affect the survival rate of CC patients. Moreover, the OS of the group combining higher CCL22+ and FOXP3+ infiltrating cells is lower than that of the group with a combined lower number of CCL22+ and FOXP3+ infiltrating cells.
In recent years, several studies have investigated the prognostic value of new immune checkpoints such as CCL22. CCL22 belongs to a family of secreted proteins that play various roles in immunoregulatory and inflammatory processes [
27]. Previous research has reported that CCL22 is a prognostic predictor of various cancers. For instance, a report by Li et al. indicated that tumor secretion by CCL22 is an independent prognostic predictor of breast cancer [
40]. Another study proved that the levels of serum macrophage-derived CCL22 are associated with glioma risk and survival period [
41]. Besides, CCL22 predicts postoperative prognosis in patients with stage II/III gastric cancer [
42]. The prognostic value of CCL22 in cervical cancer has, however, not been investigated. This study reveals that the CCL22
high group had a lower OS, compared to the CCL22
low group (
p = 0.001). CCL22
high was an independent predictor of shorter OS (HR, 4.985;
p = 0.0001). The OS of the combination group CCL22
highFOXP3
high was significantly lower than that of the combination group CCL22
lowFOXP3
low regardless of the FIGO stage and disease subtype. CCL22
highFOXP3
high was an independent indictor of shorter OS (HR, 5.284;
p = 0.009). The PFS of group CCL22
highFOXP3
high was significantly lower than that of group CCL22
lowFOXP3
low in cervical adenocarcinoma, but CCL22
highFOXP3
high was not an independent indicator (HR, 3.018;
p = 0.068). Therefore, the CCL22+ infiltrating cells or the combination of CCL22 (+) and FOXP3 (+) cells are novel biomarkers that can be potentially used for cervical cancer prognosis. Since this study only had 5/230 samples with both a high number of CCL22+ and FOXP3+ infiltrating cells, increasing the sample size might provide a more reliable result.
Previous studies reported that CCL22 could recruit FOXP3 (+) regulatory T-cells. Increased CCL22 mRNA levels are correlated with increased FoxP3 mRNA levels in oral cancer specimens [
43]. In a B16F10 melanoma model, imiquimod (IQM) has been shown to reduce T-regs at the tumor site by the downregulation of CCL22 production [
44]. In cervical cancer, the
CCL22 mRNA levels of neoplastic foci and tumor periphery is positively correlated with FOXP3 [
45]. In cervical cancer, however, the correlation between protein expression levels of CCL22 and FOXP3, and the cell source of CCL22 have not been determined. This study revealed a similar trend, that the protein level of CCL22 from both cervical cancer cells and infiltrating cells was positively correlated with FOXP3. This correlation was defined by calculating the IRS of CCL22 expression in cervical cancer cells and counting the number of CCL22+ and FOXP3+ cells. The association of the latter was better than the former. Therefore, both cervical cancer cell-derived and infiltrating cell-derived cells might recruit T-regs, and the role of CCL22 from infiltrating cells was found to be more significant than that of CCL22 from cervical cancer cells.
Several studies have investigated the source of CCL22. For instance, Huang et al. proved that CCL22 was overexpressed in head and neck cancer cells [
43]. CCL22 was found in macrophages of tongue squamous cell carcinoma [
46]. Additionally, the secretion of CCL22 by M2-like macrophages in colorectal cancer was proven [
32]. However, CCL22 expression in cervical cancer has not been previously determined. By performing a tissue microarray (TMA) using immunohistochemistry (IHC), we showed that CCL22 could be secreted by cervical cancer cells. Besides, double immunofluorescence of CD163 and CCL22 showed that 82.55% (±22.23%) of CD163(+) cells were overlapped with CCL22 (+) infiltrating cells. Therefore, CCL22 is secreted by M2-like macrophages and cervical cancer cells in cervical cancer. Moreover, many studies proved that cervical cancer cells could induce monocytes into M2 macrophages [
18,
42] which was consistent with the result in our study to some degree that CCL22 in monocytes could be induced by cervical cancer cells. However, further studies need to be performed to show the effect of cervical cancer cells on M2 macrophages and the regulatory mechanisms of CCL22 needs to be studied in the future. In conclusion, our findings suggest that the increase in CCL22 (+) infiltrating M2-like macrophage cells may recruit more T-regs in CC tissue and cause a poor prognosis for cervical cancer patients. CCL22 could be a prognostic predictor and therapeutic target to identify and treat cervical cancer patients with poorer clinical outcomes.