Association of Circulating Circular RNAs (mmu-015947, hg38-0008980, and DLGAP4) in Diagnosis, Diseases Severity, and Prognosis of Ischemic Stroke

Backgrounds Quickly diagnosing ischemic stroke (IS) is a critical issue in clinical studies, as it allows more effective therapy and stops the progression of IS. The blood level of CircRNAs after stroke may act as a rapid diagnostic marker. Methods and Results In this study, the blood level of circRNAs was evaluated using a real-time PCR. We used logistic and linear regression analysis to evaluate the potential of circRNAs levels with the risk of IS. Circ-mmu-015947 was up-regulated in patients. Its expression also showed a good value in predicting IS risk, and a signicant diagnostic value. CircDLGAP4 was decreased in patients compared with controls and logistic regression showed its expression negatively associated with IS risk. The expression level of hg38 was reduced signicantly in patients with small vessel disease (SVD) and the linear regression analysis showed a negative relationship between hg38 expression with SVD subtype. Circhg38 expression relative to controls showed a signicant association with IS risk. Conclusion Taken together, circulating circ-mmu-015947 may serve as a novel biomarker for IS, and we found a signicant decrease in the level of hg38-0008980 after IS it may act as a novel circRNA in IS pathophysiology with a positive correlation with stroke severity. correlation also that the hg38 level was signicantly correlated NIHSS scores (r = 0.19; 95% CI: -0.041 to 0.409, p = 0.04) 3. A). We did not nd signicant correlation between mmu and DLGAP4 expression level with NIHSS score. Spearman test showed a signicant negative relationship between DLGAP4 level and BMI (r = - 0.21; 95% CI: -0.422 to 0.0243, p = 0.03) in IS patients (Fig. 3. B). Our results revealed the positive correlation between the level of DLGAP4 and HDL in IS patients (r = 0.23; 95% CI: -0.003 to 0.440, p = 0.02) (Fig. 3C).


Introduction
Stroke is the most common cause of morbidity, mortality, and long-term disability in developed and developing countries [1,2]. There are several mimics and camou ages for ischemic stroke (IS) [3]. Thus, nding the rapid diagnostic test to con rm the clinical diagnosis and more effective therapy of ischemic stroke is necessary [4].
One of the endogenous single-stranded RNAs with a closed covalent-circle structure is circular RNAs (circRNAs) [5]. They are produced during post-transcriptional modi cation by back-splicing of mRNAs or lncRNAs and because of their covalent structure as well as resistance to exonuclease are much more stable than linear RNA thus can serve as a diagnostic marker in the peripheral blood of patients [6]. CircRNAs can regulate the expression of speci c target genes.
CircRNAs may also act as a diagnostic biomarker for discriminating neurological disorders [11] while they mainly have a sponge function for the regulation of microRNA (miRNA) [12]. Animal studies have revealed a close association of cerebral circRNA expression with IS [7,13,14].
Bioinformatics analysis by Lin et al. revealed that the up-regulated expression of mmu-015947 in HT22 cells with oxygenglucose deprivation/reoxygenation (OGD/R) could interact with miRNAs and may participate in metabolism, apoptosis, and immune-related pathways. According to the important roles of these pathways in the pathogenesis of IR, Lin et al. suggested that the up-regulation of mmu-015947 might be involved in the process of cerebral IRI [7]. Therefore, we presumed that the circ-mmu-015947 may function as a signi cant diagnostic biomarker in IS patients. The circDLGAP4 is a sponge for miR-143 and in animal models has been shown that the overexpression of circDLGAP4 attenuated the neurological de cits, the infarct areas, and blood-brain damage [8]. Bai et al. has shown that circDLGAP4 levels were signi cantly decreased in the acute IS patients. They demonstrated that circDLGAP4 could inhibit miR-143 activity [8].
The circDLGAP4-miR-143 network by exerting certain important regulatory effects on the permeability of the blood-brain barrier (BBB) attenuates IS outcome [16,17].
Another circRNA that has shown the important regulatory effect on BBB integrity is hg38-circ-0008980. It is a novel circRNAs could sponge hsa-miR-660-5p and exert regulatory effects of meningitic E. coli penetration on BBB [18].
According to the above-mentioned circ hg38-0008980 is likely to have a regulatory effect on BBB permeability in IS. There is no evidence about hg38-0008980 expression after IS. Thus we hypothesized that the expression of circ hg38-0008980 may alter after IS and serve as a signi cant diagnostic marker for IS. Therefore, we evaluated for the rst time the expression levels of two novel circRNAs (mmu-015947, hg38-0008980 ) and also circDLGAP4 level in peripheral blood of 75 IS patients and 75 controls by real-time PCR, to determine their expression levels within the rst 24 hours after IS to correlate these levels with the stroke severity, clinical parameters as well as their possible application for diagnosis and prognosis of IS.

Methods And Materials
Study subjects This is a case-control study conducted at Namazi Hospital in Shiraz. All patients underwent brain diffusion-weighted magnetic resonance imaging (MRI) or non-contrast computed tomography (CT), to differentiate IS from intracerebral hemorrhage. According to the Guidelines, stroke is an acute neurologic disorder lasting more than 24 hours (or < 24 hours with neuroimaging evidence) [19]. In this study, inclusion criteria were as follows: patients must be admitted within 24 h from the onset of stroke without a history of the previous stroke, older than 18 years, and we could follow them regularly.
We excluded patients who suffered from different types of cancer, treated with an immunosuppressant, severe in ammation, and patients who were pregnant or breastfeeding, patients with the transient ischemic attack (TIA) were excluded. 75 controls from Shiraz's population randomly were included in our study while were matched for sex and age with cases. The ages of the cases and controls ranged from 30-95 years. Controls with a history of stroke, TIA, or severe brain disorders were excluded. Figure 1 shows the criteria for screening, exclusion and enrolment of the study participants.
In this study, hypertension and diabetes were diagnosed according to these criteria: for hypertension: receiving antihypertensive therapy or blood pressure ≥ 140/90 mmHg on two occasions at least 24 h apart [20]. Diabetes mellitus: receiving treatment with hypoglycemic drugs or two fasting glucose >126 mg/dl (7.0 mmol/L) and 2-h post-load glucose >200 mg/dl (11.1 mmol/L) or [21]. To indicate the severity of stroke we used the National Institutes of Health Stroke Scale (NIHSS) score on admission, higher scores represent greater severity [22]. Six months after admission the functional outcomes were obtained according to the modi ed Rankin scale (mRS) blinded to the level of CircRNA [23]. Different types of IS were diagnosed according to the TOAST classi cation [24]. Ethics approval for this study was obtained by the local ethics committee of the Arsanjan Branch, Islamic Azad University, Iran (IR.IAU.A.REC.1399.029). Peripheral venous blood samples were collected after taking written informed consent from all subjects (or their proxy respondents).

Measurement of the circRNAs levels
TRIzol Reagent (GeneAll, Seoul, South Korea) was used for the isolation of total RNA from blood samples according to the thiocyanate-phenol-chloroform method. Using cDNA synthesis Kit (Yektatajhiz, Iran) for cDNA synthesis from RNA samples with the A260/A230 and A260/A280 ratios greater than 1.7. We used quantitative real-time PCR for mmu-015947, circDLGAP4, and hg38-0008980 levels measurement using RealQ Plus 2x Master Mix Green Low ROX™ (Ampliqon, Denmark). TBP was used as an internal reference for circRNAs detection. The Quantstudio 3 Real-Time PCR System (Applied Biosystems, Foster City, USA) was used while the thermal-cycling settings were 1 repeat (10 min at 95 C) accompanied by 40 cycles (15s at 95 C, 30s at 62 C, 45s at 72 C) and 1 cycle (5min 72 C). The melting phase was (15s at 95 C, 30s at 72 C, 15s at 95 C). Variation in expression levels was expressed using the cycle threshold (Ct) values. ΔCt represents the difference of Ct between TBP and the target gene. The relative circRNAs expression levels for every individual were de ned using 2 −ΔCt . The results were calculated with the 2 −ΔCt method [25].

Statistical analysis
The circRNAs expression levels were shown as mean ± SE. The following tests were used for different analyses in our study, The student's t-test for comparing the mean expression of the circ-RNAs between case and control groups. One-Way ANOVA for comparing the circRNA expression levels between different types of stroke. A comparison between categorical data was carried out using a chi-square test. The expression level of circRNAs was compared between cases and controls using an independent two-sample t-test. We used logistic regression analyses to evaluate the association of circRNA expression with the risk of IS. relationship between circRNA levels with clinical parameters was analyzed using linear regression. By ROC curve analysis the diagnostic and prognostic values were estimated and the ndings were presented as the area under the curve (AUC). Correlations were analyzed using the Spearman correlation. The analyses were done using the SPSS software (version 19.0) and GraphPad Prism 5.0. The p-value of < 0.05 was regarded as statistically signi cant.

Demographic and clinical characteristics of all participants (IS patients and controls)
The control group was similar in age, sex, and BMI to the case group (Table 1). Risk factors such as diabetes, hypertension, and smoking were demonstrated to be more prevalent in cases compared to controls. In laboratory ndings, there were no signi cant differences in the levels of triglyceride and total cholesterol between the IS cases and the controls; however, low-density lipoprotein was signi cantly higher and high-density lipoprotein levels were signi cantly lower in IS cases compared to controls (Table 1). Expression of mmu-015947 in IS patients was higher than controls In IS patients, mmu expression was shown higher than the controls at 0-24 hours after stroke (2.04 ± 0.12 vs 0.97 ± 0.09) (p < 0.0001, Fig. 2A). Furthermore, we studied and compared the mmu-015947 expression level in 4 subtypes of IS, results showed the level of mmu-015947 in SVD (p < 0.0001), LAA (p < 0.0001), and UD IS (p < 0.01) were signi cantly higher than those in the controls. But no marked differences were found between the levels of mmu-015947 in the 4 subgroups F (4, 145) = 11.48, p < 0.0001 (Fig. 2B). Moreover, logistic regression analysis was performed to test the association of CircRNA expression with the risk of IS. mmu level was signi cantly associated with risk of stroke (OR = 1.008, 95% CI: 1.78-4.19, p < 0.0001) after adjusting for age, gender, and other risk factors. Hypertension was also signi cantly showed a positive association with the risk of stroke (OR =1.54, 95% CI: 1.74-12.5, p < 0.002) (Data not shown).
Expression of hg38-0008980 and circDLGAP4 in IS patients were lower than controls.
The blood levels of CircRNAs DLGAP4 and hg38 in IS patients were signi cantly lower than the controls at 0-24 h after stroke respectively [(1.90 ± 0.11 vs 2.23 ± 0.10) and (1.71 ± 0.13 vs 2.37 ± 0.24)] (p < 0.05, Fig. 2A). Furthermore, One-Way ANOVA analysis showed the level of DLGAP4 and hg38 in SVD cases were signi cantly lower than those in the controls (p Association between clinical variables with CircRNAs (mmu-015947, hg38-0008980, and DLGAP4) expression The association between clinical variables and the circRNAs expression was de ned by subgroup analyses ( Table 2). The levels of mmu expression were shown to be signi cantly higher in the positive hypertension subgroup (p < 0.01) and also revealed that the level of hg38 expression signi cantly was higher in patients with NIHSS scores > 7. Furthermore, we also used linear regression analysis to detect the association between the circRNA levels with clinical parameters and types of stroke. We found the signi cant negative relationships between hg38 expression and SVD subtype (p = 0.008, Beta = -0.036) and also between DLGAP4 level and BMI in IS patients (p = 0.02, Beta = -0.29) ( Table 3).  (Fig. 3. A). We did not nd signi cant correlation between mmu and DLGAP4 expression level with NIHSS score. Spearman test showed a signi cant negative relationship between DLGAP4 level and BMI (r = -0.21; 95% CI: -0.422 to 0.0243, p = 0.03) in IS patients (Fig. 3. B). Our results revealed the positive correlation between the level of DLGAP4 and HDL in IS patients (r = 0.23; 95% CI: -0.003 to 0.440, p = 0.02) (Fig. 3C).
Diagnostic and prognostic value of circulating circRNAs (mmu-015947, hg38-0008980, and DLGAP4) in IS ROC curve analysis showed the signi cant potential of mmu expression for diagnosing IS cases from the controls with an AUC of 0.775 ± 0.039 (95% CI: 0.697-0.853, p < 0.0001) and the sensitivity and speci city were 80% and 72%, respectively (Fig. 4A). The expression level of DLGAP4 and hg38 in 0-24 h after stroke also showed a nonsigni cant diagnostic value for discriminating IS patients from the controls with AUC of 0.57 and 0.59 respectively ( Fig. 4B and C). The sensitivity and speci city were 78.67% and 38.67% for diagnostic value of DLGAP4 level and 56%, and 60%, for hg38 (p > 0.05).
In our study, we considered the mRS score of 3-6 in 6 months after stroke as an unfavorable functional outcome and mRS score of 0-2 as a favorable outcome. The mmu, DLGAP4, and hg38 levels showed no signi cant predictive prognosis for a 6-month unfavorable outcome relative to a favorable outcome with an AUC of 0.55, 0.58, and 0.55 respectively. These results demonstrated the expression levels of three circular RNA could not be identi ed as a functional outcome prediction marker for IS. The sensitivity and speci city for mmu, DLGAP4, and hg38 were as follows (86.54% and 30.43%), (80.77% and 39.13%), and (71.15% and 47.83%) respectively (Data not shown).

Discussion
There are still no clinical studies that identify circRNAs mmu-015947 and hg38-0008980 as biomarkers for diagnosis or progression of IS, therefore, our research was the rst study that assessed circmmu and hg38 expression 0-24 hours after IS which showed that circ-mmu was up-regulated in IS patients and had a good value in predicting IS risk, circhg38 was decreased in patients compared with controls and logistic regression showed a signi cant association of its expression with IS risk. Moreover, we found a positive correlation between hg38 level and NIHSS. The expression level of hg38 was reduced signi cantly in peripheral blood of patients with SVD and the linear regression analysis showed a negative relationship between hg38 expression with SVD subtype.
CircDLGAP4 was decreased in IS patients compared with controls and its expression was negatively associated with IS risk. CircDLGAP4 had a negative correlation with BMI (r = -0.2, p = 0.03) and positive correlation with HDL (r = 0.23, p = 0.02).
Dharap et al reported miR-329 could inhibit in ammation and neuronal death by suppressing the expression of peroxisome proliferator-activated receptor-gamma (PPARγ) [29]. Therefore it seems that the elevated mmu-015947 expression may function as a neuroin ammation enhancer in IS pathophysiology. We found high expression of mmu-circRNA-015947 in peripheral blood of 75 IS patients 0-24 h after stroke relative to controls. According to our results elevated mmu expression can increase the risk of IS (OR =1.008, 95% CI: 1.78-4.19, p < 0.0001) and it also showed a signi cant diagnostic value with an AUC of 0.77. In patients with hypertension, the mmu expression showed signi cant elevation compared with cases without hypertension while linear regression analysis did not show the signi cant relationships between mmu expression and hypertension. Hypertension is a risk factor for atherosclerotic diseases and IS [30]. Signi cant upregulation of mmu-015947 in IS patients with hypertension may increase the susceptibility to IS. In our study, there was no signi cant difference between the mean NIHSS score of patients with hypertension and patients without hypertension (9.77 ± 1.03 vs 9.78 ± 1.06) respectively. We did not nd a signi cant Spearman correlation between mmu expression and NIHSS score in patients with hypertension and also in patients without hypertension (Data not shown). We found a non-signi cant positive correlation between mmu expression and stroke severity (r = 0.01, p = 0.9).
Probably, larger sample size may be needed to show the signi cant correlation between mmu level and NIHSS score.
In our study, a signi cant decrease in the expression level of circDLGAP4 was seen 0-24 h after IS. This result is consistent with previous studies that demonstrated the downregulation of circDLGAP4 and its protective effect in IS patients [8,31]. We could not nd the negative correlation between DLGAP4 expression and NIHSS score but logistic regression analysis showed the negative association between DLGAP4 level with risk of IS. This result was con rmed by the signi cant negative and positive correlation between BMI and HDL with DLGAP4 level respectively. Because the decrease in BMI and increase in HDL level decrease the risk of IS [32,33]. Our result showed patients with lower BMI and higher HDL relative to other patients had elevated DLGAP4 expression with a lower risk of IS.
In the zho study, the mean NIHSS score was 7.9 ± 3.4, while the mean NIHSS score in our patients was 9.8 ± 0.74. Therefore, the value of circDLGAP4 might not be well correlated with NIHSS in patients with severe IS. 75 IS patients were enrolled in our study it seems that to reach a signi cant correlation between gene expression with stroke severity a larger sample is needed. The expression of circDLGAP4 showed a signi cant decrease relative to controls in our patients with SVD type. Cerebral small vessel disease has a signi cant relationship with type 2 diabetes mellitus [34,35]. A previous study showed that circDLGAP4 promotes diabetic kidney disease injury by sponging miR-143 [36]. According to the abovementioned, there is a possibility of a correlation between DLGAP4-miR-143 and diabetes as well as SVD.
Circ-hg38-0008980 expression showed signi cant downregulation 0-24 h after stroke while logistic regression analysis represented a negative association with risk of IS. Subgroup analysis and Spearman test showed a signi cant positive correlation between Circ-hg38-0008980 level and NIHSS score moreover, linear regression has also shown the negative relationship between hg38 level and SVD stroke that was con rmed by the signi cant decrease in hg38 expression level relative to controls. We also found the lower mean NIHSS score in SVD patients relative to the others (7.96 ± 1.09 vs 9.8 ± 0.74) respectively. Therefore, it seems that hg38-0008980 level as a novel circRNA may have a positive correlation with stroke severity especially in patients with SVD.

Conclusions
This was the rst study revealing the signi cant increase in the expression level of circ-mmu-015947 and a signi cant decrease in hg38-0008980 expression in IS patients. Our results showed a signi cant positive correlation of hg38 with NIHSS and a signi cant diagnostic value of the circ-mmu-015947 expression after stroke. Larger sample size with more information needs to reach appropriate statistical power. A small number of IS patients with CE stroke had enrolled in our study; therefore, in this study, the value of circRNAs in patients with CE might not be well evaluated. To further clarify the role of circulating circular RNAs (mmu-015947, hg38-0008980, and DLGAP4) in IS, additional research with a larger sample size is needed to study the diagnostic and prognostic value as well as downstream effects of these circRNAs. Further researches are necessary to nding the time-dependent changes in expression levels of these circRNAs after IS and their correlation with stroke severity. Taken together, circulating circ-mmu-015947 could serve as a novel biomarker for the diagnosis of IS, and the hg38-0008980 level showed a positive correlation with stroke severity. We found also the downregulation of circDLGAP4 after stroke while its expression showed a negative association with the risk of IS.