1. Introduction

The occurrence of renal allograft acute rejection (RAAR) has decreased greatly from 40% to 10% in the past ten years in United States [1]. RAAR remains, however, a main factor that triggers chronic rejection and graft failure [1,2]. Early detection of RAAR is critical for treatment and affects the rate of long term graft survival [3]. Techniques for noninvasive and early diagnosis of RAAR are thus needed to minimize late treatment and reduce percutaneous biopsies.

Serum creatinine (sCr) is a clinically used biomarker for indicating potential RAAR occurrence. Patients with 25% or higher elevation of sCr levels are expected to develop kidney dysfunction and usually undergo a biopsy for final diagnosis [4]. However, sCr is not an optimal biomarker despite its clinical reliance due to the well-known limitations for early detection and accurate indication of kidney dysfunction. For example, at the first week after the transplantation surgery, sCr does not reflect kidney dysfunction because the patient does not reach a steady state [5]. During patient recovery from transplant surgery, the generation and excretion rates of creatinine are not equal. Under this condition, the continuous rise of sCr does not necessarily correlate with allograft dysfunction, but instead indicates that the levels of sCr are not steady. This uncertainty potentially increases the risk of late treatment for RAAR, which significantly decreases the success of the graft [6]. Even at steady state levels, creatinine generation, excretion and volume distribution in serum are regulated by many factors other than RAAR [5]. For example, edema, a common post-transplant symptom, would dilute sCr and hence delay recognition of renal dysfunction.

In this paper, we report the use of SERS for early RAAR and renal dysfunction diagnosis. SERS is a surface-sensitive technique realized by the localized surface plasmon resonance (LSPR) a of nanostructured noble metal under laser excitation. Within the LSPR range (usually less than 10 nm from the surface of noble metal nanostructure) [7], Raman scattering cross sections can be enhanced up to 10¹⁴ fold, which enables its ultra-high sensitivity [8–10]. In addition to high sensitivity, SERS offers many other advantages over competing sensing techniques for urinary biomarkers, such as mass spectrometry, Elisa, PCR. One advantage is that the processing time of the assay is very short (less than 1 min). Also, SERS involves minimum sample preparation. For measurements in this paper, urine samples were tested as received without any further preparation. We have previously reported our initial exploratory work using a SERS-active fiber probe to analyze urine samples that illustrated the promise of the SERS method [11]. The spectral quality and measurement sensitivity were hampered by the strong Raman background of the fiber material itself. In the contrast, work presented in this paper was done using a silicon substrate immobilized with Ag nanoparticles (Ag NPs), which provided significantly higher SERS sensitivity and specificity for predicting acute kidney dysfunction. To the best of our knowledge, this is the first reported use of high sensitivity SERS for post-transplant kidney dysfunction diagnosis. Our results suggest that SERS may be useful as a clinical technique for early detection of RAAR.

2. Materials and methods

2.1. Urine samples

Urine samples were collected at Newark Beth Israel Medical Center (NBIMC) after obtaining written informed consent from 58 patients who were followed for six months after the transplant. More specifically, urine samples were collected at six time intervals: the day of, one day after, one week after, one month after, three months after, and six months after the transplant. These time intervals were labeled as A, B, C, D, E and F respectively. Patients’ urine samples were de-identified and labeled from 1 to 58. Each sample was labeled as a combination of number and letter. For instance, sample 23B represents the urine sample from patient number 23 collected one day after the transplant. All samples were immediately aliquoted into 1.5 mL centrifuge tubes and stored in −80°C freezer until further analysis. Based on the clinical biopsy results that were provided by the participating physicians at NBIMC, patients 24, 26, and 27 were diagnosed as RAAR. Patient 24 was diagnosed one month after transplant and patients 26 and 27 three months after. Two other patients, 3 and 10 were diagnosed as graft failure one week and three months after transplant, respectively. Table 1 shows the detailed clinical diagnosis based on allograft biopsy.

Table 1. Clinical Diagnosis of Patients Entered into This Study.

View Table

2.2 Synthesis of silver nanoparticles and fabrication of SERS-active substrates

Ag NPs were chosen to enhance Raman signals for the SERS measurements. Negatively charged Ag NP colloidal solution was prepared following the modified Lee and Meisel method [12] by sodium citrate reduction of silver nitrile under UV radiation. Briefly, 0.8 mL of 1% wt. sodium citrate solution was added drop-wise to 40 mL of 1 mM silver nitrile solution. The mixture was then placed in a UV chamber for 4 hours under continuous stirring. To avoid excessive temperature increase due to UV irradiation, a water batch was used to keep the temperature under 50 °C to ensure monodispersity of Ag NPs. The Ag NPs have an average size of 40 ± 5 nm, a ζ-potential of −40 mV, and a plasmon resonance peak of 406 nm. Since the surface of a silicon wafer with a thin native oxide is also negatively charged, polyallylamine hydrochloride (PAH) (Mw 1500g, Aldrich) was introduced as an anchoring layer to provide positive substrate surface charge for immobilization of Ag NPs via electrostatic attraction. Here, silicon substrates were initially immersed in PAH aqueous solution (0.2 mg/mL, pH 9) for 20 minutes followed by rinsing in Milli-Q water at pH 4.5 gently. Si substrates functionalized with PAH were immersed in Ag NP colloidal solution (10¹⁶ particle/mL) for 8 hours in the dark to allow surface attachment of Ag NPs. The resultant substrates were then carefully rinsed in Milli-Q water at pH 4.5 and used immediately for SERS measurements.

2.3 Raman spectroscopy

A custom-built Raman spectrometer illustrated in Fig. 1 was used for SERS analysis. Initially 532 nm laser excitation was attempted because it is closer to the resonance wavelength of the Ag NPs. However, it induced a significant florescence intensity that overwhelmed the Raman spectral features. 632.8 nm laser excitation proved to be effective to suppress florescence while producing high SERS signal intensity. The optic train began as a continuous He-Ne laser (632.8 nm), collimated to a single mode fiber. The laser beam then passed through a narrow band filter and was reflected by a dichroic mirror to the objective. The beam was focused at the surface of the SERS-active silicon substrate onto which each urine sample was added. The reflected light was collected by the same objective and then passed through dichroic mirror and a long-pass filter. The scattered signal was collimated to a multimode fiber and was delivered to an Acton SpectraPro 2300 spectrometer equipped with a Roper Scientific liquid nitrogen-cooled CCD camera. All SERS measurements were done at room temperature. The laser power was approximately 5 mW with an exposure time of 20 sec. Each sample was repeated 3 times at 3 different spots to ensure repeatability. Spectra were processed using OriginPro 8.5.

 

Fig. 1 Schematic of the SERS optical set up.

Download Full Size | PDF

3. Results and discussion

SERS spectra for each single patient at different time intervals were firstly compared. For most patients, urine sample B, collected one day after the kidney transplant surgery, possessed significantly richer spectral features while spectra at other time intervals appeared very similar. Figure 2 shows the urine spectra of patient 4. 520 cm⁻¹ is the characteristic vibrational band of single crystal silicon. It could be used as reference point both for Raman shift calibration and peak intensity analysis. 1005 cm⁻¹ is the primary Raman peak of urea, the dominant nitrogen-containing component in urine [13]. It corresponds to the symmetrical C-N stretch [14] and is present in all urine spectra of patient 4. In spectrum 4B, 1360 cm⁻¹ appeared as the primary peak and was almost four times higher in intensity than the silicon peak. Peaks such as 427, 662, 779, 1443 and 1550 cm⁻¹ were only observed in 4B but absent in all other time intervals. We thus focused on all samples collected at time B as (a) spectra at time B contain the most distinctive information compared to other time intervals and (b) such information may be informative for timely diagnosis of kidney dysfunction.

 

Fig. 2 Urine spectra of patient 4 at time intervals A, B, D and E. BG represents the background spectrum of Ag NPs on the silicon substrate. Each spectrum was shifted vertically for clarity in illustration.

Download Full Size | PDF

By investigating all urine spectra at time interval B, unique spectral features were noticed in all five patients who prospectively developed kidney dysfunction. Figure 3 shows the comparison of spectra of these patients with those from five normal subjects. The spectra of three RAAR patients, 24, 26, and 27, displayed very similar features except for an additional peak at 1005 cm⁻¹ for 27B. All three RAAR patients had an intense peak at 1360 cm⁻¹ along with two peripheral peaks at 1448 and 1495 cm⁻¹. The 1360 cm⁻¹ peak ranged from 5~15 times higher than the silicon peak at 520 cm⁻¹. Patient 3, who was diagnosed as graft failure one week after transplant, shows a very unique and extensive spectral feature where the 1005 cm⁻¹ peak became dominant and highly intense, approximately 8-fold higher than the silicon peak. The other graft failed episode, patient 10, displayed a very similar spectrum as the three AR episodes. In spectra of normal patients, the intensity of 1360 cm⁻¹ peak was lower than the silicon peak at 520 cm⁻¹ or not present and peripheral 1448 and 1495 cm⁻¹ peaks were even weaker.

 

Fig. 3 (a) SERS spectra of the urine samples from RAAR patients at time interval B. (b) SERS spectra of the urine samples from normal subjects at time interval B.

Download Full Size | PDF

To numerically analyze the spectral features further, the 1360 cm⁻¹ peak was used as the critical feature of kidney dysfunction and the ratio of intensities of 1360 over 520 cm⁻¹ was calculated for each spectrum at time interval B. Here, the silicon peak was taken as the reference for intensity analysis of the 1360 cm⁻¹ peak since it remains fairly constant from sample to sample. Each spectrum was analyzed using LabSpec software (version 5.58). Areas under the 520 and 1360 cm⁻¹ peaks were integrated. The baseline was automatically truncated in the integration so that only peaks were considered. A peak width for integration was used as 20 and 70 cm⁻¹ for 520 and 1360 cm⁻¹, respectively. ‘R’ was used as the abbreviation of ratio of intensity of the 1360 cm⁻¹ over the 520 cm⁻¹ peak. Figure 4 shows a comparison of SERS-based analysis versus a clinical diagnosis of kidney dysfunction. All three RAAR patients have relatively high R values. For patients 24, 26 and 27, the R values are 15.35, 9.94 and 5.53 respectively. Patient 10 was diagnosed as acute graft failure 3 weeks after transplant and the R value was 7.49. Patient 3 is the other acute kidney failure episode diagnosed one week after transplant and the R value was 2.39. Patient 24 had the highest R (15.35) of all 58 patients. The distribution of R values of patients is visualized in Fig. 4(a), with clinical diagnosis shown in different colors. Only three patients (24, 41 and 17) had R larger than 10. Eight patients had R values between 5 and 10. Eleven patients have R values larger than 5, including all kidney dysfunction patients but patient 3. Thus, an intense 1360 cm⁻¹ peak (R larger than 5) indicated a significantly high potential for acute kidney dysfunction. Patient 3 did not have intense 1360 cm⁻¹ peak and only two other patients, 5 and 49, shared similar spectral patterns. To summarize, urine samples of 14 patients exhibited significant distinctive SERS spectral features one day after transplant surgery and that correlated a high risk of acute kidney dysfunction occurrence. None of the 7 normal patients in groups R>10 and 10>R>5, including patient 17 and 41 who had comparably high R value, showed abnormity or received biopsy based on available clinical diagnosis of transplant patients up to 6 months post operation. It is not clear why their R values are high. We continue to follow these patients to see if the R values might serve as an indicator for chronic rejection or long term kidney allograft survival. On the other hand, 8 patients who received biopsy are in the R<1 group based on the SERS spectral analysis. Clinically, 18 patients received biopsies indicated by high sCr levels and acute rejection and graft failure were diagnosed ranging from one week to 3 months after transplant (Fig. 4(b)). However, none of the 15 normal patients who received biopsy had R values larger than 5. All in all, the R value does not appear to have a strong correlation with sCr levels. Overall, our studies indicate that SERS spectral features are slightly more specific to acute kidney dysfunction than sCr.

 

Fig. 4 (a) Distribution of the intensity ratio of peaks at 1360 cm⁻¹ over 520 cm⁻¹. Three patients had R values higher than 10, including patient 24 (RAAR episode). Eight patients had R values between 5 and 10, including patients 26, 27 (RAAR episodes), and patient 10 (acute kidney failure). The R values of twelve patients are between 1 and 5, including patient 3 (acute kidney failure). The R values of the other 37 patients were less than 1. (b) Results from clinical diagnosis. Eighteen patients received biopsy due to high levels of sCr. Five patients were diagnosed as acute rejection or kidney failure. The remaining 40 patients were considered normal and did not receive a biopsy.

Download Full Size | PDF

With similar specificity with respect to predicting kidney dysfunction as the current clinical gold standard, sCr, urine SERS analysis provides a much earlier indication, only one day after the transplant surgery. Clinically, sCr indicates abnormal renal function one week after the transplant for patient 3; one month later for patient 27; 3 months after transplant for patients 10, 24 and 26. Hence, SERS analysis of urine, with only one day post-transplant indication of renal abnormal function, provides a significantly earlier prognostic indicator permitting patients to be treated earlier to increase the graft survival rate.

These special SERS spectral features appeared only one day after transplant in patients’ urine samples, A literature search was conducted to identify the possible analytes that caused the spectral features that we observed. The top candidates were heme, a ferrous contained porphyrin molecule in human hemoglobin and myoglobin. The major Raman peak for heme agrees with our observations. It has been reported that intense peak at 1356~1376 cm⁻¹ has been observed in Raman spectra of human hemoglobin and myoglobin under various excitation [15,16]. The intense peak at 1356~1376 cm⁻¹ originates from the pyrrole half-ring symmetrical stretch within the heme molecule. Also, the presence of heme in kidney transplant patients’ urine is very common. Hematuria, a common symptom after a kidney transplant especially at early days following surgery, may be a potential source of heme in urine. It has been reported that patients with persistent hematuria after renal transplant have higher potential to have urologic malignancy [17,18] and higher possibility of having an acute rejection [17]. However, hemoglobin, or hematuria, is not the only form of heme. Free heme molecules are more likely to offer the intense 1360 cm⁻¹ peak. Firstly, free heme molecules provide more intense Raman signal for SERS sensing than heme wrapped in hemoglobin because heme may reside nearer to a hot spot between Ag NPs without the separation of protein coating since the gross shape of hemoglobin is an ellipsoid of 25 × 30 × 40 Å in dimension. That dimension excludes the 27alanines and 20 glycines occurring on the surface, considering which the overall size is bigger. Heme is buried in a hydrophobic cavity in the center of hemoglobin [19]. Moreover, heme embedded in hemoglobin should be separated further from Ag NPs if the protein is confined in red blood cells, which have dimensions about 7~8 µm in diameter. Second, free heme may cause severe hemolysis, lipid peroxidation, DNA damage, protein damage and aggregation [20–24]. In the kidney specifically, heme may lead to renal failure by stimulating local inflammatory responses [25]. Thus, we hypnotize that free heme is more likely to cause the SERS peak at 1360 cm⁻¹ than hemoglobin protein in urine.

Pure heme molecule is neither stable nor commercially available. To ascertain the above hypothesis, SERS measurement of protoporphyrin IX (P-IX) (95%, Aldrich) was conducted under the same experimental condition as urine samples. P-IX is the biological precursor of heme and has similar structure of heme molecule but lacks an iron atom in the middle of the ring (Fig. 5).

 

Fig. 5 Chemical structure of (a) protoporphyrin IX and (b) heme B.

Download Full Size | PDF

SERS measurements of P-IX was carried out using an aqueous solution at 10⁻⁷ M. This concentration was chosen in recognition of the fact that heme, if any, likely exists only in minute amount in the urine of transplant patients. Shown in Fig. 6 is the SERS spectrum of the P-IX solution, with a strong band at 1350 cm⁻¹. In light of the fact that Raman band may shift, depending on the orientation of adsorbed molecules on the Ag surface of the SERS substrate, this band is sufficiently close to the 1360 cm⁻¹ band observed in the urine samples of RAAR patients. This result lends significant support to our postulation that free heme may be the most probable source of the prognostic Raman feature for RAAR and it can hence be used as the potential urinary biomarker for early indication of RAAR.

 

Fig. 6 SERS spectrum of protoporphyrin IX (P-IX) in aqueous solution at 10⁻⁷M. BG represents the background spectrum of Ag NPs on the silicon substrate.

Download Full Size | PDF

4. Conclusion

Urine analysis using SERS has been shown to have a great potential for early indication of acute kidney dysfunction. As early as one day following transplant surgery, urinary SERS spectra show distinctive peak features at 1360 cm⁻¹, which correlate high risk of acute kidney dysfunction. With this information, patients can be treated earlier following this timely diagnosis so that appropriate clinical invention can be pursued and hence the graft survival rate can be raised significantly. Moreover, SERS analysis is more specific to renal abnormal function than sCr, which may lower the usage of biopsies, reducing patient discomfort and medical costs. Free heme is hypothesized to be the major biomarker for special SERS feature at 1360 cm⁻¹. Given that a portable Raman system is already commercially available, SERS-based analysis holds excellent promise as a bedside clinical tool for acute kidney dysfunction indication.