Long Noncoding RNA HOTAIRM1 Promotes Immunosuppression in Sepsis by Inducing T Cell Exhaustion

Sepsis, which remains the leading cause of death in intensive care units (1), is an acute life-threatening disorder associated with multiorgan dysfunction resulting from a dysregulated host immune response against pathogen infections (2). Sepsis-induced lung injury, as respiratory tract infections are the most common site of infection, is associated with the highest sepsis-related mortality (3). In the early state of sepsis, tissue injury or bacterial infection can trigger a defense reaction of the body, inducing excessive activation of immune cells and the release of proinflammatory mediators. The late phase of sepsis, however, alters innate and adaptive immune responses with long-term immunosuppression, which is accompanied by persistent inflammation and enables the development of persistent, recurrent, secondary, and nosocomial infections. The vast majority of patients with sepsis survive the initial insult. However, increasing evidence shows that a longer period of immunosuppression is the key to poorer outcomes and increased long-term mortality (4). To date, no Food and Drug Administration–approved treatment methods have been shown to improve sepsis survival, although >100 therapeutic clinical trials on sepsis have been performed (5). Thus, there is an urgent need to explore and identify the underlying cause of long-term immunosuppression and develop a targeted therapy for treating the late phase of sepsis.

Sepsis-induced immunosuppression is multifactorial and features defects in both the innate and adaptive immune systems. For one thing, apoptotic depletion of immune effector cells in sepsis has been reported, including NK cells, CD4⁺ and CD8⁺ T cells, B cells, and dendritic cells (6, 7), mediated by death receptor– and mitochondrial-mediated pathways (8). The apoptotic cells, together with and their damaging impact on surviving immune cells, lead to massive immunosuppression in late sepsis (9). In addition, excessive regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs) decreased expression of positive costimulatory molecules, and the increased expression of negative costimulatory molecules also contributed to sepsis-induced immunosuppression (10). Furthermore, T cell exhaustion characterized by increased PD-1 and excessive Tregs is closely associated with the mortality of sepsis (11).

Long noncoding RNAs (lncRNAs), transcripts longer than 200 nt, regulate the expression levels of target genes through multiple mechanisms and play roles as competing endogenous RNAs to determine the fate of gene transcripts (12). HOTAIRM1 (HOXA transcript antisense RNA myeloid-specific 1) is a conserved lncRNA involved in myeloid differentiation that has been found to be dysregulated in multiple cancers. It promotes cancer development mainly by regulating cancer cell differentiation and migration (13). However, its role in immune regulation has just started to be discussed in recent studies. It has been found that HOTAIRM1 could promote the expansion and immunosuppressive function of MDSCs (14); however, whether it participates in T cell exhaustion remains unclear.

Using a mouse model of sepsis, together with both in vivo and in vitro studies, we found that the high expression of HOTAIRM1 was critical for sepsis-induced immunosuppression. First, the expression of HOTAIRM1 was upregulated by Notch/Hes1 activation. Blockade studies further demonstrated that HOTAIRM1 was required for the induction of PD-1⁺ T cells and Tregs, marked as T cell exhaustion. Mechanistically, HOTAIRM1 upregulated PD-L1 expression by targeting HOXA1. Taken together, our data implicated the pathogenic role of HOTAIRM1 and the Notch/Hes1/HOTAIRM1/HOXA1/PD-L1 axis in sepsis-induced immunosuppression.

Three septic patients diagnosed in accordance with the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) (15) and admitted to intensive care units were recruited from Zhongshan Hospital, Fudan University (Shanghai, China). None of these patients received neoadjuvant therapy. Peripheral blood was collected in sampling tubes containing EDTA or heparin at day 6–8 after the onset of sepsis, considered as late sepsis with immunosuppression (16). Three healthy volunteers served as healthy controls. All participants provided written informed consent. This study was approved by the Ethics Committee of Zhongshan Hospital, Fudan University (B2021-182R), and complied with the ethical standards set out in the Declaration of Helsinki.

C57BL/6J male mice (6–8 wk old) were purchased from Shanghai Laboratory Animal Research Center (Shanghai, China) and maintained by the Department of Laboratory Animal Science of Fudan University under a 12-h light/12-h dark cycle and specific pathogen-free conditions. Mice were housed in groups of five per cage and acclimated for a week before the start of experiments. All animal experiments were conducted in accordance with the relevant guidelines and regulations approved by the Animal Ethics Committee of Zhongshan Hospital, Fudan University.

Sepsis was induced by cecal ligation and puncture (CLP) as described previously (17). Briefly, mice were anesthetized by i.p. injection of 1% pentobarbital sodium (1 mg/kg). A 1- to 2-cm-long abdominal incision was made and the cecum was ligated distal to the ileocecal valve, punctured twice with a 23-gauge needle. Once needle-sized feces were detected, the cecum was placed back to its original position and the abdomen was closed. All animals received 0.5 ml/10 g of normal saline for rehydration. Early and late sepsis approximating the clinical condition of human sepsis were established as described in previous studies (18, 19). Briefly, mice were s.c. administered antibiotics (imipenem; 25 mg/kg body weight) or an equivalent volume of 0.9% saline at 8 and 16 h after CLP. These levels of injury and manipulation create prolonged infections with high mortality (∼60–70%) during the late phase. Early sepsis is defined as the first 5 d after CLP whereas late sepsis occurs after day 5 (20). Survival was followed for 28 d. Sham mice underwent the same surgical procedures without cecal puncture or ligation. In some experiments, mice were then intranasally inoculated with 20 μl of Pseudomonas aeruginosa containing 1 × 10⁶ CFU of bacteria for a second insult at 7 d after CLP.

To knock down HOTAIRM1, mice were sedated with 5% isoflurane, and adeno-associated virus (AAV)-6-sh-HOTAIRM1 (obtained from Genelily BioTech) was endotracheally administered at a multiplicity of infection of 10⁶ to the lungs through an endotracheal tube inserted into the trachea. Then, 3 wk later, mice were used for sepsis induction, following the same procedure for wild-type mice. To investigate the effects of Notch1 and PD-L1 on the progression of sepsis, a Notch1 inhibitor (FLI-06, 1 mg/kg) or PD-L1 inhibitor (atezolizumab, 10 mg/kg) was diluted in 100 μl of 0.9% normal saline and injected i.v. at day 0 (early sepsis) or day 5 (late sepsis) after CLP with the effective concentration used in murine studies (21, 22).

For further analysis, lung tissue was collected and fixed with 4% paraformaldehyde at 4°C or immediately stored at −80°C. Whole blood was centrifuged at 3000 rpm for 15 min, and serum was collected and stored at −80°C. Bronchoalveolar lavage fluid (BALF) was intubated three times using a total volume of 1.5 ml of sterile PBS and centrifuged at 4°C at 1500 rpm for 5 min.

Blood samples were collected from mouse cheeks, and murine PBMCs were isolated by density gradient cell separation using Histopaque-1083 (Sigma-Aldrich, St. Louis, MO). Separated PBMCs were treated with RBC lysis buffer containing 139.5 mM NH₄Cl and 1.7 mM Tris-HCl (pH 7.65) at 37°C for 10 min and were then washed with 0.1% BSA/PBS. Murine PBMCs were incubated with an Ab mixture for 30 min at 4°C after treatment with a mouse Fc blocker to block nonspecific binding sites. The stained cells were analyzed using FACSVerse. The proportion of the designated cell fraction was determined by recording 10,000 events, and data files were analyzed using FlowJo software (Tree Star, Ashland, OR). The following Abs were used: PE anti-mouse CD4 (catalog no. 100407, BioLegend), BV510 anti-mouse CD8 (catalog no. 126631, BioLegend), FITC anti-mouse CD25 (catalog no. 101907, BioLegend), allophycocyanin anti-mouse CD62L (catalog no. 104427, BioLegend), BV650 anti-mouse CD127 (catalog no. 135043, BioLegend), and PerCP-Cy5.5 anti-mouse PD-1 (catalog no. 135223, BioLegend).

Three pairs of blood samples of sepsis patients and healthy volunteers were used for lncRNA microarray analysis. GeneChip Operating Software (Affymetrix, Santa Clara, CA, USA) was used to analyze differentially expressed lncRNAs. Fold change > 2.0 and p < 0.05 were the criteria for selecting lncRNA for subsequent analysis.

Total RNA was extracted from lung tissues by TRIzol (Invitrogen, USA), and cDNA was extracted by an ABI high-capacity cDNA reverse transcription kit (Thermo Fisher Scientific, Waltham, MA). Quantitative real-time PCR was performed using the ABI StepOnePlus real-time PCR system (Applied Biosystems, Foster City, CA), with internal reference to GAPDH. The primers used are shown in Table I, and the results were calculated using 2^−ΔΔCT.

Total RNA was extracted from lung tissues by the standard TRIzol method, and electrophoresis was performed via formaldehyde-denatured agarose gel electrophoresis. Isolated RNA was then transferred to a nitrocellulose membrane. The membrane was incubated with a mixed buffer for 2 h. The membrane was then incubated with a biotin-labeled RNA probe. Enzyme-labeled streptavidin (Thermo Fisher) was used to detect the biotin signal according to the manufacturer’s instructions.

Whole blood was collected into a syringe through heart puncture, and the sample was placed in a test tube at room temperature for 30 min. After centrifugation (6000 rpm, 15 min, 4°C), the plasma and supernatant were immediately frozen in liquid nitrogen (LN2) and stored at −80°C for further analysis. The levels of IL-10 and TGF-β in plasma were measured with an ELISA kit (Boster Biological Technology, Wuhan, China). Cytokine and chemokine concentrations (CXCL1, GM-CSF, TNF-α, IL-6) were determined in BALF using commercial ELISA kits (Boster Biological Technology). The results were spectrophotometrically determined using a bacteriophage reader.

HOTAIRM1 expression was detected in paraffin-embedded lung tissues by an in situ hybridization kit (Boster Biological Technology). Briefly, the specimens were deparaffinized, rehydrated, and then incubated with proteases at 37°C for 10 min. After washing three times with PBS, the specimens were incubated overnight at 40°C. Then, blockade was performed with blocking buffer for 30 min and anti-digoxigenin (ab76907, Abcam, Cambridge, UK) for 60 min. Finally, a diaminobenzidine kit (Solarbio, Beijing, China) was used for staining. Positive signals are shown in red.

The right lung was separated, weighed (wet weight), and then dried overnight in an oven at 60°C (dry weight). The wet weight/dry weight ratio was calculated by dividing the wet weight by the dry weight.

Right lung tissue was removed, washed with PBS, fixed with 4% PFA, and embedded in paraffin. For histological examination, formalin-fixed paraffin-embedded lung tissue was cut into 4-μm sections, placed on glass slides, dewaxed with xylene, and rehydrated with 30–100% ethanol. The sections were microwave boiled in extract buffer (10 mM citric acid buffer, pH 6.0) for 5 min and stained with H&E, and a histopathological examination was performed. The stained sections were analyzed under a light microscope (Carl Zeiss, Jena, Germany).

Western blotting was performed as described in our previous study (23). Briefly, protein was isolated from liver tissues or cells, separated by 12% SDS-PAGE, and transferred to polyvinylidene difluoride membranes. Membranes were then probed with the following primary Abs: anti–PD-L1 (ab205921, Abcam, Cambridge, UK), anti-HOXA1 (ab230513, Abcam, Cambridge, UK), and anti-GAPDH (ab8245, Abcam, Cambridge, UK).

The interactions between HOTAIRM1 and HOXA1 and between HES1 and the HOTAIRM1 promoter were predicted using the starBase version 2.0 database (http://starbase.sysu.edu.cn/) and catRAPID database (http://service.tartaglialab.com) using the sequence of HOTAIRM1 (NR_038366.1). Transcription factor prediction was performed by browsing the JASPAR database (http://jaspar.genereg.net/) and TRANSFAC database (http://gene-regulation.com/pub/databases.html) using the sequence of HOTAIRM1 (NR_038366.1).

RNA-binding protein immunoprecipitation (RIP) was performed using the Magna RIP kit (Millipore). Transfected cells were collected and lysed using RIP lysis buffer. Cell extracts were mixed with magnetic beads and incubated with a primary Ab at 4°C for 2 h. RNA was purified by TRIzol LS (Life Technology) and quantitatively analyzed by real-time PCR.

MLE-12 cells overexpressed mouse HOXA1 after transfection with pcDNA-HOXA1 or pcDNA-NC. Cells were collected 48 h after transfection. After lysis, magnetic beads coated with streptavidin were incubated with the cell lysates for 3 h at 4°C to obtain biotin-conjugated RNA complexes. Quantification of the RNA-binding protein mixture was achieved by Western blot analysis.

The chromatin immunoprecipitation (ChIP) assay was performed using an EZ-ChIP kit (Millipore, Germany). MLE-12 cells were fixed with 1% formaldehyde for 10 min, and then glycine was added to stop the fixation. Next, lysate buffer was incubated on ice with protease inhibitor mixture II for 30 min. The cells were then treated with ultrasound and centrifuged. The supernatant was collected, and protein G-agarose was added at 4°C for 1 h. A Hes1 Ab and a polymerase II Ab were added to the supernatant. Overnight incubation was then performed at 4°C. Protein was digested with Proteinase K, and chromatin was extracted and quantified by real-time PCR.

Luciferase reporter plasmids carrying the wild-type or mutant HOTAIRM1 promoter were cotransfected with the control vector or HES1 overexpression plasmid into MLE-12 cells. After 48 h, the cells were detected by a dual-luciferase reporter gene assay system (Promega). Normalization of firefly luciferase activity was based on Renilla luciferase activity.

Total protein concentration was measured in BALF using the Dc (detergent-compatible) protein assay (Bio-Rad), according to the manufacturer’s instructions, with BSA, fraction V (Roche Diagnostics, Mannheim, Germany) as a standard.

Primary alveolar epithelial cells were collected from mice lungs following a previously described procedure (24, 25). Briefly, lungs of mice were perfused with PBS, the trachea was cannulated, and 2 ml of dispase solution (Corning Life Sciences) was applied into the airways. The lungs were then excised and incubated with dispase, after which the lungs were chopped and the bronchi were removed. Dissected tissue was then incubated at 37°C for 45 min and the obtained cell suspension was sequentially filtered through 100- and 40-μm cell filters and centrifuged. After digestion with DNase, the collected primary murine epithelial cells were purified by differential centrifugation and immune adherence (IgG) and cultured with DMEM plus 10% FBS.

For the proliferation assay, CD4⁺ and CD8⁺ T cells were labeled with CFSE (2 μM) for 10 min. Cells were then activated by anti-CD3/anti-CD28 and stimulated with PMA (50 nmol/L) and ionomycin (500 nmol/L) at 37°C, 5% CO₂ for 48 h. Cells were then harvested and CFSE was detected by flow cytometry.

For inducible IFN-γ production of T cells, cells were stimulated with PMA (50 nmol/L) and ionomycin (500 nmol/L) for 4–6 h at 37°C, 5% CO₂, in the presence of 1 mg/ml brefeldin A (BD Biosciences). Surface staining of CD4/CD8 was performed, and cells were fixed and permeabilized and stained with intracellular IFN-γ. The percentage of IFN-γ–positive cells in CD4⁺ and CD8⁺ T cells was detected by flow cytometry.

In vivo pulmonary bacterial clearance was determined as previously described (26). In brief, mice were inoculated intranasally with 5 × 10⁶ CFU of Escherichia coli in a 50-μl volume. At 6 h postinfection, lungs were removed, homogenized, serially diluted, and plated on trypticase soy agar. After overnight incubation, manual plate counts were performed to quantify bacterial burden in the lung.

For assessment of the BALF antimicrobial activity, 90 μl of BALF and 10 μl (5 × 10³ CFU) of the E. coli suspension were mixed into wells of a 96-well plate and then incubated at 37°C with constant agitation for 90 min. Bacteria were disaggregated by vigorous discharge through a pipette, diluted, and plated on trypticase soy agar plates. After overnight incubation of the plates at 37°C, CFU were counted.

Murine BALF cells were identified by the profile of Ab staining as alveolar macrophages (CD45⁺CD11c⁺F4/80⁺), lymphocytes (CD45⁺CD3⁺), neutrophils (CD45⁺Ly6G⁺), and monocytes (CD45⁺CD11b⁺F4/80⁺). Absolute BALF leukocyte numbers were calculated by multiplying total BALF cell counts by the percentage of cells in each subpopulation as determined by flow cytometry.

The experimental data were statistically analyzed by SPSS 17.0 software. All experiments were repeated at least three times independently. Quantitative data are expressed as the mean ± SEM. The mean of two independent samples was compared by a Student t test, and the mean of multiple groups was compared by one-way ANOVA. A p value <0.05 was considered statistically significant. One-way ANOVA was used for multi-group comparisons followed by Bonferroni post hoc analysis. Survival was determined by Kaplan–Meier survival curve and analyzed using log-rank test.

To explore whether HOTAIRM1 is involved in the development of immunosuppression associated with the late phase of sepsis, blood samples from late sepsis patients and healthy controls were collected for microarray analysis. Highly expressed lncRNAs (fold change > 2.0), including HOTAIRM1, were identified in blood samples from late sepsis patients, compared with those from healthy controls (Fig. 1A, Table I). To verify the expression profile of HOTAIRM1, a murine CLP sepsis model was established. The early phase of sepsis is defined as the first 5 d after CLP whereas late sepsis refers to 5 d after CLP (18, 20). Blood samples and lung tissues were collected from either early or late sepsis mice for real-time PCR and Northern blot analysis. Consistent with the clinical data, significantly upregulated expression of HOTAIRM1 was found in the serum of late sepsis mice, compared with that of early sepsis (Fig. 1B). As expected, increased HOTAIRM1 expression was also confirmed in the lung tissues of late sepsis mice (Fig. 1C, 1D). Anti-inflammatory cytokines IL-10 and TGF-β were elevated in late sepsis (Fig. 1E) and used to assess the immunosuppressive phase (19). Thus, to further explore the role of HOTAIRM1 in the development, especially the immunosuppression, of late sepsis, the correlation between serum IL-10/TGF-β and the expression of HOTAIRM1 in late sepsis mice was analyzed. (Fig. 1F suggested the trends for a correlation between the serum and lung tissue HOTAIRM1 expression and the level of IL-10 and TGF-β from late sepsis mice. Therefore, it is found that the expression of HOTAIRM1 increased in late sepsis and could be associated with sepsis-induced immunosuppression in mice.

On this basis, we further investigated how HOTAIRM1 contributes to the progression of immunosuppression in late sepsis. To this end, sh-HOTAIRM1 transfection was used to knock down HOTAIRM1 in mice, whereas transfection of AAV-NC was used as the negative control. As shown, higher HOTAIRM1 levels were found in both serum and lung tissues from the untreated and AAV-NC–transfected late sepsis mice, but they were not observed in the sh-HOTAIRM1–transfected group (Fig. 2A, 2B), suggesting the depletion of HOTAIRM1 by sh-HOTAIRM1. In situ hybridization further verified the significantly increased expression of HOTAIRM1 in the lung tissues of late sepsis mice, as well as its efficient knock down by sh-HOTAIRM1 transfection (Fig. 2C). Then, survival rate and lung injury in wild-type and HOTAIRM1-depleted mice were evaluated to study the contribution of upregulated HOTAIRM1 to disease progression. We found that HOTAIRM1 depletion significantly decreased the mortality in late sepsis (Fig. 2D). Moreover, lung injury in late sepsis was ameliorated in HOTAIRM1-depleted mice (Fig. 2E), which was further indicated by a reduced wet weight/dry weight ratio of lung tissues (Fig. 2F), the lower total cell number in BALF (Fig. 2G), and the lower total protein concentration in the BALF (Fig. 2H). In summary, we found that depleting HOTAIRM1 could decrease mortality and reduce lung injury in late sepsis mice, therefore suggesting that increased HOTAIRM1 expression contributed to the progression of late sepsis.

Next, to determine how HOTAIRM1 promotes lung injury and intensifies the progression of late sepsis, we assessed the T cell spectrum, which includes the primary pathogenic immune cell in sepsis. The percentages of total CD4⁺/CD8⁺ T cells, naive CD4⁺/CD8⁺ T cells, PD-1⁺ CD4⁺/CD8⁺ T cells, and Tregs were evaluated by flow cytometry analysis (Fig. 3A). Compared to early sepsis, late sepsis mice showed significantly reduced percentages of both total and naive CD4⁺/CD8⁺ T cells in the lungs and spleens and slight differences in the lymph nodes, which were reversed by HOTAIRM1 depletion (Fig. 3B, 3C, Supplemental Figs. 1A, 1B, 2A, 2B). However, immunosuppressive Tregs and PD-1⁺ CD4⁺/CD8⁺ T cells increased in late sepsis, which were also inhibited by depleting HOTAIRM1 (Fig. 3B–D, Supplemental Fig. 1A–C). As for proliferative function, CFSE staining was applied. After activation and stimulation, both CD4⁺ and CD8⁺ T cells from late sepsis mice showed a lower CFSE-positive percentage, suggesting slower division and proliferation rates, which, however, were mildly reversed by HOTAIRM1 depletion (Fig. 3E, Supplemental Fig. 1D). The impaired production of IFN-γ upon stimulation in CD4⁺ and CD8⁺ T cells in late sepsis mice were also found to be reversed by depleting HOTAIRM (Fig. 3F, Supplemental Fig. 1E). Altogether, these data indicate the induction of T cell exhaustion in HOTAIRM1-depleted late sepsis mice. As PD-L1/PD-1 signaling is a well-known inhibitory signaling pathway dominating the immunosuppression and exhaustion of T cells, the expression of PD-L1 was then measured in mice with late sepsis. Compared to early sepsis, the mRNA and protein levels of PD-L1 were found significantly increased in the lung tissues of late sepsis mice, while they were reduced by the knockdown of HOTAIRM1 (Fig. 4A, 4B), which may contribute to the expansion of PD-1⁺ T cells. Overall, the change of T cell subsets suggested that HOTAIRM1 could promote the immunosuppression in late sepsis by inducing T cell exhaustion.

To further investigate how HOTAIRM1 induces T cell exhaustion, primary alveolar epithelial cells isolated from mice (Supplemental Fig. 2C) were used for HOTAIRM1-binding target screening. Through a starBase prediction and RNA pull-down, HOXA1 was identified as a HOTAIRM1 binding factor in alveolar epithelial cells (Fig. 4C). Further pull-down assays using murine alveolar epithelial MLE-12 cells confirmed the binding of HOTAIRM1 to HOXA1 (Fig. 4D, 4E), which was then verified by their colocalization, as indicated by immunofluorescence staining (Fig. 4F). In this case, we further investigated whether the transcription factor HOXA1 is responsible for the upregulation of PD-L1 mediated by HOTAIRM1. HOTAIRM1 and HOXA1 were knocked down in MLE-12 cells by small interfering RNA transfection, respectively (Fig. 4G). A dual-luciferase assay (Fig. 4H) was then conducted, showing that both HOTAIRM1 and HOXA1 knockdown impaired PD-L1 expression in MLE-12 cells, indicating that HOTAIRM1 induced the expression of PD-L1 by targeting the transcription factor HOXA1. Overall, our results demonstrated that HOTAIRM1 could contribute to immunosuppression in late sepsis by promoting PD-L1/PD-1 inhibitory signaling and subsequently T cell exhaustion in late sepsis mice.

Currently, the upstream regulator of HOTAIRM1 is unknown. By browsing the Promo database and TRANSFAC database, we identified a promising binding site of Hes1 at the promoter of HOTAIRM1 (Fig. 5A). To confirm whether Hes1 is a key factor regulating HOTAIRM1 expression, a luciferase reporter system was developed by inserting the HOTAIRM1 promoter into the pGL3 vector, using a mutant prompter as a negative control (Fig. 5B). We found significantly upregulated HOTAIRM1 when Hes1 was overexpressed (Fig. 5C). ChIP assays further confirmed the direct binding of Hes1 and HOTAIRM1 (Fig. 5D). Moreover, significantly higher Hes1 expression was found in late sepsis mice, compared with early sepsis mice (Fig. 5E). The correlation of HOTAIRM1 and Hes1 was found in the lung tissue of late sepsis mice (Fig. 5F), which further verified the role of Hes1 in regulating HOTAIRM1 expression in late sepsis.

Hes1 has been demonstrated to function as a canonical effector of Notch signaling (27). We then investigated whether Notch signaling is involved in the regulation of HOTAIRM1 in late sepsis mice and contributed to disease progression. The Notch inhibitor FLI-06 was injected at day 0 or day 5 after CLP to interfere with the early or late phase of sepsis, and lung tissues were then harvested. Significantly higher PD-L1 and HOTAIRM1 expression was found in mice of late sepsis, compared with that of early sepsis, which was inhibited upon FLI-06 treatment (Fig. 6A, 6B). The same expression pattern of PD-L1 in lung tissue was verified by immunohistochemistry (IHC) analysis (Fig. 6C). The survival rate and lung injury were then compared in mice with or without FLI-06 treatment. As expected, a significantly higher mortality was observed in late sepsis mice compared with that in the early phase, whereas FLI-06 treatment improved their survival without inducing toxic effects (Fig. 6D). Clearly ameliorated lung injury was also observed (Fig. 6E), as evidenced by the reduced wet weight/dry weight ratio of lung tissue (Fig. 6F), reduced total cell number in BALF (Fig. 6G), and lower total protein concentration in the BALF (Fig. 6H) in late sepsis mice receiving the FLI-06 treatment. Overall, our findings suggested that the Notch/Hes1 activation was responsible for the upregulated HOTAIRM1 and the subsequent immunosuppression in late sepsis.

To evaluate whether blocking the PD-1/PD-L1 inhibitory signaling would attenuate disease progression in mice with late sepsis, the PD-L1 inhibitor atezolizumab was applied. PD-L1 expression in lung tissues with or without atezolizumab treatment was detected by real-time PCR, Western blotting, and IHC analysis. As expected, significantly higher PD-L1 expression was found in late sepsis mice than in early sepsis mice (Fig. 7A, 7B). As a FcγR binding-deficient, fully humanized IgG1 mono-Ab designed to interfere with the binding of PD-L1 to its receptors (28), atezolizumab did not affect PD-L1 expression at either the mRNA or protein level in the lung tissues of late sepsis mice (Fig. 7A–C). However, it significantly improved the survival of mice with late sepsis (Fig. 7D). Consistently, lung injury in late sepsis mice was ameliorated by atezolizumab (Fig. 7E), which was further indicated by the reduced wet weight/dry weight ratio of lung tissue (Fig. 7F), total cell number in BALF (Fig. 7G), and total protein concentration in BALF (Fig. 7H). Therefore, these data revealed that anti–PD-L1 treatment could ameliorate HOTAIRM1-mediated lung injury and promote the survival of late sepsis mice.

To further determine the impact of HOTAIRM1 in post-sepsis immune suppression and the susceptibility to secondary infection, mice were subjected to P. aeruginosa 7 d after CLP. The expression of HOTAIRM1 and PD-L1 in this model showed the same increasing trend as shown in the late sepsis mice with CLP only. Similarly, the inhibitory effect of HOTAIRM1 depletion on the expression of PD-L1, as well as the decreased HOTAIRM1 by Notch inhibition, was observed in this model (Fig. 8A–C). Although the secondary P. aeruginosa pneumonia resulted in higher mortality (Fig. 8E) and induced more severe lung injury, compared with the CLP mice without a second insult, the therapeutic effects of HOTAIRM1 depletion, Notch inhibition, and anti–PD-L1 treatment were maintained (Fig. 8D, 8F–H). As suggested by the results from the CLP model, this therapeutic effect could have resulted from inhibiting immune suppression. Therefore, whether HOTAIRM1 depletion, Notch inhibition, and anti-PD-L1 treatment could lower the susceptibility to secondary infection was evaluated. The capacity of in vivo pulmonary bacterial clearance and BALF antimicrobial activity were compared. As presented, mice with HOTAIRM1 depletion, Notch inhibition, or anti–PD-L1 treatment displayed better bacterial clearance and antimicrobial activity (Fig. 8I, 8J). Notably, a higher level of inflammatory cytokines in the BALF was observed in all groups of mice with interventions, suggesting the presence of more inflammation (Fig. 8K). Furthermore, as noted by the enhanced level of CXCL1 recruiting leukocytes, accompanied by the increased number of macrophages, lymphocytes, neutrophils, and monocytes in the BALF (Fig. 8L), it is suggested that intervening the HOTAIRM1 pathway rendered sepsis mice a stronger immune response to secondary infection. Thus, these data further provide hints for the inhibition of immune suppression of sepsis by HOTAIRM1 depletion.

Altogether, we identified HOTAIRM1 as the pathogenic lncRNA advancing T cell exhaustion-mediated immunosuppression during late sepsis in the current study. For one thing, Notch/Hes1-mediated increased expression of HOTAIRM1 induced the expansion of exhausted PD-1⁺ T cells and Tregs. For another thing, PD-L1 expression was elevated in lung epithelial cells, depending on the upregulation of HOXA1 targeted by HOTAIRM1. As a result, the highly expressed HOTAIRM1 promoted the development of a T cell exhaustion-mediated immunosuppressive microenvironment in late sepsis (Fig. 9). Therefore, our findings demonstrated that the Notch/Hes1/HOTAIRM1/HOXA1/PD-L1 axis played a pivotal role in the progression of sepsis-induced immunosuppression.

Despite decades of research, sepsis remains the leading cause of death in intensive care units, characterized by severe life-threatening multiple organ dysfunction. To date, both dysregulated proinflammatory and anti-inflammatory functions have been acknowledged to trigger the onset and progression of sepsis (29). Despite that the hyperinflammatory phase (characterized by shock, fever, and hypermetabolism) initiated by the early host immune response is still the main cause of rapid death of sepsis patients (30), nowadays well-developed treatment strategies allow most patients to survive this initial hyperinflammatory phase (31). However, for those who suffer from long-term/persistent sepsis (32), the lethality of the immunosuppression phase (leading to immune dysfunction and impaired antimicrobial immunity) becomes the critical problem that needs to be conquered. Therefore, there is an urgent need to investigate and reveal the regulatory network of immunosuppression during sepsis, from which potential strategies for sepsis treatment could be explored and developed to regulate immune dysfunction caused by sepsis-induced immunosuppression.

To this end, our study aimed to reveal a part of the regulatory network in the setting of lncRNA-dominating mechanisms, which involves the regulation of a large scale of gene expression and subsequently various cellular activities. From the microarray data of late sepsis patients, several differentially expressed lncRNAs were identified. Among these, the significantly upregulated HOTAIRM1 has caught attention. Well known as a cancer-promoting lncRNA, the role of HOTAIRM1 primarily involves regulating cell differentiation, proliferation, and migration (33), whereas its function in immune regulation is largely unknown. Nevertheless, the involvement of HOTAIRM1 in the regulation of immune cells was first and only described in MDSCs so far (34). More importantly, recent studies have observed the high expression of HOTAIRM1 in sepsis (14), consistent with our result, and hepatitis C virus infection (35) as well. In both of the studies, HOTAIRM1 was related to the enhanced immunosuppression of MDSCs. According to our preliminary results and previous studies, there is reasonable evidence to hypothesize that the lncRNA HOTAIRM1 plays a pivotal role in the progression of sepsis by regulating immunosuppression. Thus, we further investigated the mechanisms involved.

Evidence for dysfunction of the immune cells that leads to immunosuppression includes 1) the massively reduced number of immune effector cells caused by apoptosis, with T cells being the most affected cell type; 2) the significantly downregulated release of proinflammatory cytokines; 3) the upregulation of PD-L1/PD-1–mediated inhibition pathways and downregulation of CD28 and HLA-DR–mediated activation pathways; and 4) the expansion of inhibitory immune cells, such as Tregs and MDSCs (36). In addition to apoptotic deletion (37), compromised T cell effector function and T cell exhaustion, as characterized by reduced production of the proinflammatory cytokines IFN-γ and TNF-α, increased PD-1 expression, and decreased CD127 expression on T cells, have also been observed in sepsis patients and are closely correlated with nosocomial infection and mortality due to sepsis (32). As we know, T cells are the primary pathogenic immune cells leading to the immunosuppression of sepsis. Therefore, we investigated the effects of HOTAIRM1 on T cells in late sepsis mice and first described the alteration of T cell subsets induced by HOTAIRM1.

We first assessed T cell number, including CD4⁺ T cells and CD8⁺ T cells, and found that HOTAIRM1 depletion significantly increased T cell number, indicating that HOTAIRM1 might be the key factor controlling T cell number in late sepsis. Further study is needed to clarify the exact mechanisms of HOTAIRM1 regulating the T cell number by assessing T cell apoptosis, proliferation, and differentiation. T cells could exhibit an impaired Ag-driven proliferative capacity and effector functions following sepsis, displaying an increased expression of inhibitor receptors such as PD-1 and a reduced ability to produce the proinflammatory cytokines IFN-γ and TNF-α, contributing to their reduced ability to prevent infection (38) and immunosuppression in late sepsis (39). The above phenomenon of T cell dysfunction induced by sepsis is called “T cell exhaustion” (11). Methods to revitalize exhausted T cells, such as blockade of the PD-1/PD-L1 interaction and IL-15 treatment, could reinvigorate immunity, alleviating immunosuppression in septic mice (40). Therefore, we explored whether HOTAIRM1 was responsible for these impaired T cell functions during late sepsis. We found that HOTAIRM1 depletion in vivo significantly reduced PD-1⁺ T cells. As mentioned, PD-1 expression is a key marker for exhausted T cells, and our results thus indicated the crucial role of HOTAIRM1 in inducing T cell exhaustion during sepsis. In future studies, apoptosis of pre-existing memory CD8⁺ T cells, Ag sensitivity and Ag-driven secondary expansion of memory CD8⁺ T cells, and Ag-independent bystander activation of memory CD8⁺ T cells in response to heterologous infection need to be detected to further clarify the mechanisms involved. Except for their regulation on exhausted T cells, HOTAIRM1 was also found to be critical for upregulating PD-L1 expression in lung epithelial cells. Altogether, HOTAIRM1 was found to be a key regulator of PD-L1/PD-1 inhibitory signaling in late sepsis.

Several genes have been identified as the downstream targets of HOTAIRM1. Its upstream regulators, however, have been hardly investigated. In the current study, we identified a promising binding site of Hes1, which functions as a canonical effector of Notch signaling, at the promoter of HOTAIRM1. Furthermore, we detected high Hes1 expression in the lung tissue of late sepsis mice and verified the interaction of HOTAIRM1 and Hes1, indicating the role of Hes1 in HOTAIRM1 expression by binding to its promoter in late sepsis. Next, through both in vitro and in vivo experiments, we verified the critical role of Notch/Hes1 signaling on controlling HOTAIRM1 expression and demonstrated that Notch/Hes1 activation was responsible for the upregulated HOTAIRM1 in late sepsis. Furthermore, Notch1 inhibition by FLI-06 attenuated the mortality and lung injury in late sepsis mice, indicating that it could be a potential target for sepsis treatment. However, increased mortality has been seen in the later time point after FLI-06 in our mouse model. Notch signaling has been known to protect CD4⁺ T cells against apoptosis in the inflammation models (41). Therefore, the phenomenon we observed could be caused by an increase in T cell death induced by Notch inhibition. In this case, a cell-specific Notch inhibiting strategy will be needed for sepsis treatment.

Overall, our study demonstrated, to our knowledge for the first time, that the Notch/Hes1/HOTAIRM1/HOXA1/PD-L1 axis contributed to sepsis-induced immunosuppression, which could be potential targets for sepsis therapies. Due to the limitation of patient sample collection in the clinic, our major findings were based on a mouse model of sepsis. However, mice used in our current model and most of the other studies were of young age (42), approximating an adolescent human being, which has the less risk of sepsis compared with the elderly. Therefore, a sepsis model with aged mice should be applied in future studies to better represent the clinical features and to extrapolate findings from the mouse model to the sepsis patients. Additionally, our future studies will focus more on identifying this Notch/Hes1/HOTAIRM1/HOXA1/PD-L1 axis in late sepsis patients and exploring potential targets for disease treatment in the clinic. In summary, we report that the lncRNA HOTAIRM1, which is modulated by Notch/Hes1 signaling, plays a crucial role in promoting immunosuppression and disease progression in late sepsis. It functions through recruiting and binding to its target HOXA1 and then further inducing PD-L1/PD-1 inhibitory signaling, which contributes to the T cell exhaustion-mediated immunosuppression in late sepsis.

The authors have no financial conflicts of interest.