Development and application of a method for detecting RNA of West Nile virus genotypes 1 and 2 based on loop-mediated isothermal amplification

Full Text

Introduction

West Nile virus (WNV, Orthoflavivirus nilense) is a member of the Flaviviridae family and belongs to the Japanese encephalitis virus group. According to the classification of pathogenic biological agents adopted in Russia, WNV is classified as a Group II pathogenicity virus. The virus is characterized by high genomic variability. WNV strains of genotypes 1 and 2 are pathogenic to humans [1].

The incidence of West Nile fever (WNF, ICD-10 code A92.3) has been officially recorded in 51 regions of the Russian Federation. Currently, the northern boundary of locally acquired WNV cases in European Russia runs through the Kostroma Region, and in Far East — through the Khanty-Mansi Autonomous District–Yugra [2]. Global warming is leading to the emergence of new territories suitable for the habitat of mosquitoes that transmit the virus, which contributes to the expansion of the WNV range [3, 4].

In this regard, improving laboratory diagnostics for WNV is a priority. Currently, serological, virological, and molecular-genetic methods are used. The virological method is labor-intensive and time-consuming, while the immunological method is associated with the possibility of false-positive results due to the presence of cross-reacting antigens in WNV and closely related orthoflaviviruses [5]. The application of nucleic acid amplification techniques makes it possible to rapidly and accurately detect WNV RNA in clinical and zoological-entomological samples, enabling epidemiological surveillance of the pathogen and facilitating early diagnosis [6].

Currently, the polymerase chain reaction (PCR) is the most widely used molecular genetic method in the diagnosis of WNV [7–9]. In Russia, four reagent kits are registered for detecting WNV RNA using reverse transcription PCR (RT-PCR). To amplify the target WNV cDNA sequence during PCR, a cyclic heating is required to denature the double strand.

In recent years, isothermal methods have been actively developed, in which the denaturation of the double-stranded DNA is achieved by the displacement activity of the polymerase. One of the most widely used technologies is loop-mediated isothermal amplification (LAMP) [10–12]. This method uses a specific Bst polymerase (DNA polymerase I from the thermophilic bacterium Geobacillus stearothermophilus), which performs cyclic DNA synthesis at a fixed temperature (60–70°C) and possesses a unique ability to pass through complex regions of cDNA — hairpins [13, 14]. To detect RNA in the analysed sample, an enzyme with reverse transcriptase activity is added to the reaction mixture, enabling reverse transcription and loop-mediated isothermal amplification to be performed in a single tube (RT-LAMP) [10, 15].

The high specificity of LAMP is ensured by selecting 4 or 6 primers that recognize 6 or 8 regions of the target sequence. During the reaction, a mixture of DNA fragments of varying lengths is formed, representing inverted repeats of the target sequence [10, 16, 17].

Despite the complexity of the molecular mechanism and the high requirements for primer selection, this method is simple to perform in practice. After adding the sample to the reaction mixture, it is incubated at 60–65°C in a thermostat, thermal cycler, or water bath. The reaction products formed in LAMP are detected visually by a color change (when using dyes, such as hydroxynaphthol blue) or by the clouding of the reaction mixture resulting from the formation of magnesium pyrophosphate [10, 18]. Possible detection methods include electrophoresis, pH measurement, the use of intercalating dyes, and fluorescently labeled primers [19–21].

According to research findings, Bst polymerase exhibits greater tolerance to reaction inhibitors compared to Taq polymerase, which is used in PCR. This allows to analyse the native material (such as plasma, cerebrospinal fluid, or swabs) or reduces the time required for nucleic acid extraction by using rapid methods, such as thermal lysis or brief microwave irradiation of whole blood [22, 23].

The ability to use rapid methods for sample preparation, reaction performance, and LAMP result detection without expensive equipment allows for on-site analysis at the point of sample collection. The main advantage of LAMP is the rapid turnaround time, which is important both in field settings and in high-tech laboratories and can be a critical factor when a rapid diagnosis is needed or when analyzing a large number of samples.

LAMP-based reagent kits have been proposed in Russian for a wide range of pathogens: Salmonella spp., Mycobacterium tuberculosis complex, Plasmodium spp., Candida auris, Aspergillus niger, measles virus, influenza B virus, and SARS-CoV-2. Foreign authors have reported positive results in the development and application of tests for detecting WNV RNA using the RT-LAMP method. In most studies, the sequence encoding protein E was used as the target during primer selection. Various methods for detecting results have been proposed: electrophoretic [24], turbidimetric [24, 25], fluorescent [25], immunochromatographic [26], and colorimetric [27]. However, no domestic LAMP test for detecting WNV RNA has yet been registered as a medical diagnostic products.

The aim of the study is to develop a method for detecting WNV RNA of genotypes 1 and 2 using the RT-LAMP method with different detection options and to validate it using on clinical material and samples collected during epizootic surveillance.

Materials and methods

Samples under study

The study utilized two WNV strains (Orthoflavivirus nilense WNV_Volg911/22, WNV_Volg601/18), two heterologous virus strains (Alphavirus sindbis SINV_Volg673/19, Orthobunyavirus ebiense EBIV_Volg308/20), one Escherichia coli ATCC 25923 strain from the collection of the Volgograd Research Institute, and two vaccine virus strains (Orthoflavivirus flavi 17D, Institute of Poliomyelitis and Viral Encephalitis behalf of MP Chumakov, Rubivirus rubellae RA-27/3, Mikrogen).

WNV-infected material: clinical (n = 8) — blood, serum, and plasma; cerebrospinal fluid; urine; postmortem (n = 13) — brain and spinal cord, kidney, liver, spleen, trachea, lung; field (n = 9) — tick and mosquito suspensions, suspensions of brain and internal organ tissues from mice and birds, mouse blood. Material infected with heterologous viruses (n = 2): a blood sample containing Orthoflavivirus denguei (type 1), a tick suspension sample containing Orthonairovirus haemorrhagiae.

All clinical samples from patients and biological (field) samples were obtained from the Reference Center for Monitoring of the West Nile Fever Causative Agent, Testing Laboratory Center (Volgograd Plague Control Research Institute).

To assess analytical sensitivity, the facility’s standard samples (FSS) of WNV cDNA (FSS cDNA WNV-1 and FSS cDNA WNV-2), containing fragments of the genome sequences of WNV genotypes 1 and 2 were used (Russian Federation Patent No. 2744096 dated April 3, 2020, Russian Federation Patent No. 2744095 dated April 3, 2020).

Sample preparation

Work with clinical (biological) samples was conducted in accordance with the requirements of SanPiN 3.3686-21 “Sanitary and Epidemiological Requirements for the Prevention of Infectious Diseases,” MU 1.3.2569-09 “Organization of the work of laboratories using nucleic acid amplification methods when working with material containing microorganisms of pathogenicity groups I–IV.”

WNV RNA isolation was performed using the RIBO-Zol-S and RIBO-Prep reagent kits (Central Research Institute for Epidemiology) in accordance with the manufacturer’s instructions. The isolated RNA was stored at –70°C prior to analysis.

Conducting RT-PCR

To confirm the presence of WNV RNA and determine the genotype, RT-PCR with hybridization-fluorescence detection was performed using the Ampligen-WNV-Genotype-1/2/4 reagent kit (RU RZN 2022/17020, Volgograd Plague Control Research Institute) on a Rotor-Gene Q (Qiagen) instrument.

Computer programs

Multiple alignment of whole-genome nucleotide sequences of WNV was performed using the Unipro UGENE v. 1.31.0 software. The selected target was compared with sequences from the GenBank genetic database to establish homology and exclude possible cross-reactions using the NCBI BLASTn algorithm. The search for and analysis of specific primers were performed using the PrimerExplorer 5 web service¹ (Eiken Chemical Co.). Using the PerlPrimer v. 1.1.21 computer program, the probability of dimer formation, hairpins, and other secondary structures of the selected primers was calculated.

Isothermal amplification with reverse transcription

RT-LAMP was performed in a single tube. The 25-μL reaction mixture contained 8 primers from the kit: F3 — 5 pmol, B3 — 2.5 pmol each, FIP — 40 pmol, BIP — 20 pmol each, LF, LB — 40 pmol each; a mixture of 2'-deoxynucleoside-5'-triphosphates (Syntol) — 1.6 mM each; buffer for Bst polymerase; 24 units of Genta Bst DNA polymerase; 75 units of Genta REV L reverse transcriptase (GenTerra); 10 μL of RNA template.

Primer synthesis was performed by Syntol. As a negative control, an equal volume of TE buffer was added to the tube in place of the sample.

LAMP with electrophoretic detection

The reaction was carried out over 45 minutes: reverse transcription at 50°C for 15 minutes, followed by amplification at 65°C for 30 minutes. Incubation was performed on a programmable thermostat “Tertzik” (DNA-Technology). LAMP products were detected in a 2% agarose gel prepared using standard methods based on Tris-borate buffer [28]. Electrophoresis was performed for 50 min at a voltage gradient of 5 V/cm. LAMP reaction products were visualized by staining with ethidium bromide solution at a concentration of 0.5 μg/mL. LAMP results were recorded under transmitted ultraviolet (UV) light using the GelDoc XR (Bio-Rad) and its accompanying QuantityOne software. The results were evaluated based on the presence of DNA fragments whose bands formed a characteristic ladder formation.

RT-LAMP with end-point fluorescence detection

For visual detection, 2 μL of the intercalating dye EtBr (0.08 ng/mL) was applied to the inner surface of the reaction tube cap prior to sample addition. Incubation was performed as in the previous method. After completion of the RT-LAMP, the tubes containing the reaction mixture were mixed to bind the dye to the amplicons. Fluorescence was detected using a UV lamp, a TCP-20 transilluminator (Vilber Lourmat Sté), and a GelDoc XR gel documentation system (Bio-Rad).

RT-LAMP with real-time fluorescent detection

For RT-LAMP, with real-time monitoring of results, staining was performed using an intercalating dye added to the reaction mixture. Two dyes were used in this study: 20x Eva488 (Lumiprobe RUS) and 50x SYBRGreen I (Eurogen).

The RT-LAMP duration was 40 minutes: reverse transcription step at 50°C for 15 minutes; amplification under isothermal conditions at 65°C for 5 cycles of 50 seconds each, without detection; followed by 25 cycles of 50 seconds each with detection on the Green (FAM) channel. Reactions were performed on Rotor-Gene Q amplifiers (Qiagen GmbH) and DTprime (DNA-Technology).

Results

To develop a LAMP-based method for detecting WNV RNA, multiple alignment was performed on the whole-genome nucleotide sequences of WNV genotypes 1, 2, and 4 circulating in Russia. The most conserved fragment of the WNV sequence of the epidemiologically significant genotypes 1 and 2 was selected as the target—the 5'-untranslated region (5'UTR) and the locus encoding the capsid protein (protC) of the polyprotein gene. WNV genotype 4 sequences were considered heterologous, given the unproven pathogenicity of this genotype for humans and its genetic heterogeneity.

Using the PrimerExplorer V5 online service, five sets of primers were selected for the reference genome of the WNV genotype 2 (NC_001563.2, strain 956), differing in their position on the target sequence. In the first stage of the study, LAMP was performed at 65°C for 60 min, and the results were evaluated by gel electrophoresis. As samples, we used FSS cDNA of WNV genotypes 1 and 2 at concentrations of 1 × 10⁴–1 × 10⁵ genome equivalents per mL (GE/mL).

Analysis using 4 out of 5 primer sets revealed nonspecific amplification in the form of a smear in the negative control on the electropherogram. In the reaction with the fifth set, WNV_LAMP_Ref, a positive result was recorded with samples of genotypes 1 and 2 of the WNV, and a clear, specific ladder formation was observed on the electropherogram. However, for WNV genotype 1 cDNA, a smaller number of amplification products was observed on the electropherogram, and when testing low concentrations of WNV genotype 1 cDNA (less than 5 × 10⁴), a negative result was obtained.

After analyzing multiple alignment of the nucleotide sequences of the WNV genotype 1, critical substitutions were identified at the binding sites of the reverse outer WNV_LAMP_Ref-B3 and reverse inner WNV_LAMP_Ref-BIP primers. Using a manual selection method, additional primers WNV_LAMP_Ref1-B3 and WNV_LAMP_Ref1-BIP were designed for the reference sequence of the WNV lineage 1 genotype (West Nile virus lineage 1, NC_009942.1), into which substitutions were introduced based on the principle of degeneracy. Loop primers (WNV_LAMP_Ref-LF, WNV_LAMP_Ref-LB) were also selected for the WNV_LAMP_Ref primer set to improve reaction efficiency. Fig. 1 displays the primer binding sites.

 

Fig. 1. A fragment of multiple alignment of nucleotide sequences of WNV strains. Sequences: WNV genotype 1 — JX041634.1, MN149538.1, AY277252.1, MZ605381.2; WNV genotype 2 — MN619801.1, OR757512.1, PQ889005.1, PQ889017.1, PQ889030.1; WNV genotype 4 — FJ159129.1, FJ159130. Sequence for primer design WNV_LAMP_Ref-F3, WNV_LAMP_Ref-B3, WNV_LAMP_Ref-FIP, WNV_LAMP_Ref-BIP, WNV_LAMP_Ref-LF, WNV_LAMP_Ref-LB — NC001563.2; for primers WNV_LAMP_Ref1-B3 and WNV_LAMP_Ref1-BIP — NC009942. 1. Dots indicate nucleotides that match the WNV genotype 2 sequence NC001563.2.

 

The inclusion of additional primers increased the sensitivity of the reaction for WNV genotype 1. The effect of loop primers on the amplification rate was evaluated by varying the incubation time. When performing LAMP for 30 and 60 minutes with loop primers, the number of specific amplification products on the electropherogram did not differ. In contrast, when using only external and internal primers, a significant decrease in reaction efficiency and a negative amplification result were observed for low cDNA concentrations as incubation time was reduced (Fig. 2).

 

Fig. 2. Electrophoregram of LAMP products of WNV cDNA at a concentration of 1 × 10⁵ GE/mL with varying incubation times. 1 — 60 min with loop primers; 2 — 30 min with loop primers; 3 — 60 min without loop primers; 4 — 30 min without loop primers; 5 — negative control.

 

When selecting the optimal polymerase with displacement activity, we compared SD Hotstart polymerase (BIORON GmbH) with Bst polymerases from various manufacturers: Bst 2.0 WarmStart, Bst 3.0 (New England Biolabs), Genta Bst DNA polymerase (GenTerra), Bst DNA Polymerase (Biolabmix). With the SD polymerase, a negative result was obtained regardless of variations in temperature, incubation time, and the presence of preheating. When using the Bst 3.0 enzyme, false-positive results were observed in the negative amplification control and a smear on the electropherogram. The other Bst DNA polymerases from various manufacturers were characterized by the absence of nonspecific amplification. It was shown that the use of the Biolabmix polymerase with the developed primer set leads to the detection of signals in later cycles and a relatively low level of fluorescence. The processivity indices of Bst 2.0 WarmStart and Genta Bst DNA polymerases were comparable and depended on the buffer composition.

During the study, the composition of the 5× LAMP buffer was optimized: 70 mM Tris-HCl (pH 8.5 at 25°C); 20 mM (NH₄)₂SO₄; 7 mM MgSO₄; 0.01% Tween-20; 0.1 mg/mL bovine serum albumin; 5% glycerol. The proposed buffer differed from the commercial one only in pH and glycerol concentration (5%), but its use resulted in more stable performance of Genta Bst DNA polymerase and a reduction in the time required for amplification products to appear.

MgSO₄ and MgCl₂ were added to the reaction mixture as sources of Mg²⁺ ions. The Mg²⁺ concentration was varied from 2 to 10 mM. It was observed that the optimal concentration of MgSO₄ for cDNA amplification of WNV was 8–10 mM. However, when performing reverse transcription and LAMP in a single tube, a decrease in reaction sensitivity was observed with increasing Mg²⁺ concentration due to inhibition of reverse transcriptase activity. Testing of enzymes from various manufacturers showed that the maximum MgSO₄ concentration at which effective cDNA synthesis occurred for MMLV reverse transcriptase (Central Research Institute of Epidemiology) and M-MuLV–RH reverse transcriptase (Biolabmix) was 5 mM, and for Genta REV M reverse transcriptase (GenTerra)—7 mM. High sensitivity in the detection of WNV RNA using Genta REV M reverse transcriptase and a buffer containing 7 mM MgSO₄ was achieved by increasing the concentration of Genta Bst polymerase to 24 units in the reaction.

In the RT-LAMP format, the reverse transcription time was varied. The optimal temperature for Genta REV M reverse transcriptase is 50°C, and the enzyme should remain active at temperatures up to 65°C. We compared RT-LAMP results obtained by incubating at 65°C for 45 minutes, as well as with preliminary incubation at 50°C for 5, 10, 15, 20, and 30 minutes, followed by 65°C for 30–40 minutes. Optimal cDNA yield in samples with low viral load was obtained by performing reverse transcription at 50°C for at least 15 min.

Visual detection of LAMP results at the endpoint using ethidium bromide dye was tested. After the reaction, a bright red-orange fluorescence was observed under UV light in positive samples, while fluorescence was minimal in negative samples (Fig. 3).

 

Fig. 3. Visual detection of LAMP amplicons under UV light. 1 — negative control (RNA buffer); 2 — Orthonairovirus haemorrhagiae RNA; 3 — human blood sample infected with WNV genotype 2; 4 — human cerebrospinal fluid sample containing WNV genotype 1; 5 — mosquito suspension containing WNV genotype 2.

 

An intercalating dye was added to the reaction mixture for RT-LAMP. The effect of two dyes on the LAMP reaction was analyzed: SYBR Green I at concentrations of 0.5×, 1×, and 2× in combination with 4% DMSO, and Eva488 at concentrations of 1×, 1.5×, and 2×. The study found that 1× and 2× SYBR Green I inhibits the LAMP reaction, and reducing its concentration lowers the fluorescence level, which complicates data interpretation. For Eva488, minor fluctuations in reaction sensitivity were observed when the dye concentration was varied (Fig. 4). Among the intercalating dyes, 1.5× Eva488 was selected as optimal, as its use ensured a stable fluorescence level at low concentrations of WNV cDNA (5 × 10³ GE/mL). The Eva488 dye did not inhibit the LAMP reaction and provided a good increase in fluorescence, which allowed for a fairly accurate interpretation of the result.

 

Fig. 4. Fluorescence curves and LAMP cycle threshold values as a function of intercalating dye concentration (Green channel). А — 1× Eva488; В — 1,5× Eva488; С — 2× Eva488; D — 0,5× SYBR Green I; E — 1× SYBR Green I; F — 2× SYBR Green I.

 

Thus, the final set of oligonucleotides included three internal and three external primers, as well as two loop primers. The optimal reaction mixture composition included enzymes manufactured by GenTerra and an experimental buffer containing 7 mM MgSO₄. Three detection methods were proposed: electrophoretic, end-point fluorescent, and real-time.

The developed method for detecting WNV RNA based on RT-LAMP with real-time detection was validated on 22 samples of clinical and autopsy material, 10 samples of zoological and entomological material, as well as 7 strains of viruses and bacteria. The analytical specificity of the reaction was 100%; when testing the samples under investigation, a positive RT-LAMP result was obtained for all samples infected with WNV. No cross-reactions with other microorganisms were detected.

Analytical sensitivity was determined by serial dilution of WNV cDNA FSS at concentrations of 1 × 10³, 5 × 10³, 1 × 10⁴, 5 × 10⁴, and 1 × 10⁵ GE/mL. The limit of detection was defined as the lowest concentration of the control sample that was positive in all replicates. The analytical sensitivity of the RT-LAMP reaction using the developed primer set was 5 × 10³ GE/mL.

Discussion

Given the high variability of viral genomes, primer selection is the most challenging and critical aspect of developing a LAMP test system. Oligonucleotides were selected based on strict criteria: high specificity toward the target gene, absence of non-specific binding sites, and minimal inter- and intra-primer interactions. Oligonucleotides forming stable hairpins and dimers were excluded. Particular attention was paid to the conservatism of the 3'-ends.

When loop primers, which anneal in the regions between the inner and outer primers, were included in the reaction mixture, an increase in the amplification rate was observed, which is consistent with literature data on the increased yield of target products due to the use of 1 or 2 loop primers [16, 29].

The choice of enzymes plays an important role in the sensitivity and specificity of the reaction. There are examples of the use of SD-polymerases in LAMP that exhibit higher thermal stability compared to Bst-polymerase, allowing for the addition of a primary high-temperature DNA denaturation step, which increases reaction efficiency [30, 31]. However, no amplification occurred in the reaction using the proposed set of primers and SD polymerase. Imported enzymes are expensive, and there are currently supply issues with these reagents. In some cases, the use of the new-generation Bst 3.0 enzyme, which possesses reverse transcriptase, displacement, and polymerase activities, resulted in nonspecific amplification in negative samples. Testing of domestically produced Bst DNA polymerases and reverse transcriptase with various buffers allowed for high RT-LAMP sensitivity.

Among the many options for evaluating LAMP results to detect WNV RNA, endpoint detection (via electrophoresis and visual assessment under UV light) and real-time fluorescent detection were tested. Electrophoresis as a method for detecting results during LAMP development is convenient for assessing reaction specificity, but its use in diagnostics is not practical due to the time required and the risk of contamination.

End-point detection with an intercalating dye has been proposed as a method for visual assessment of LAMP results, which can be recommended for point-of-care diagnostics in resource-limited settings. Implementing this approach requires a minimal set of technical equipment, such as a thermostat and a portable fluorometer or UV lamp [32]. The prospects for using LAMP to detect WNV RNA in the field have been demonstrated [25, 27].

One of the most accurate and convenient methods for detecting amplification products is real-time fluorescent detection. Intercalating dyes are most commonly added to the RT-LAMP reaction mixture for this purpose. The main requirement for these dyes is that they have no inhibitory effect on Bst-DNA polymerase activity at working concentrations. According to the literature, the optimal dyes are SYTO-9, SYTO-82, EvaGreen, SYBRGreen, and berberine [33–36]. Reaction setups using SYBRGreen I and Eva488 with the developed primer set showed that the Eva488 dye had a lesser effect on the efficiency of WNV cDNA amplification.

The detection limit for WNV RNA using RT-LAMP, according to various authors, can range from 0.1 WNV PFU/mL to 10² copies of synthesized RNA/μL [24, 26]. Validation of the developed primer set for the detection of WNV RNA of genotypes 1 and 2 by RT-LAMP on samples of clinical and zoological-entomological material, as well as heterologous strains of viruses and bacteria, demonstrated high sensitivity and specificity. The analytical sensitivity of the reaction (5 × 10³ GE/mL) is comparable to that of registered reagent kits for detecting WNV RNA by RT-PCR. Increased amplification specificity is achieved by using a larger number of primers compared to PCR, which helps to avoid false-positive results.

Conclusion

During this study, a set of LAMP primers specific to a conserved region of the genome of epidemiologically significant WNV genotypes was designed. The composition of the reaction mixture and the parameters for reverse transcription and isothermal amplification were optimized, taking into account the thermodynamic characteristics of the reaction. Various methods for detecting the results were tested. Visual end-point detection is convenient for analysis using a thermostat and a portable fluorometer in the field. RT-LAMP requires PCR equipment; however, when using the Eva488 dye, the developed method demonstrated high analytical performance (reaction sensitivity was 5 × 10³ GE/mL, specificity — 100%), while the time to obtain results was reduced to 1 hour, which is at least twice as fast as PCR. The effectiveness of RT-LAMP for detecting WNV genotypes 1 and 2 in samples collected during epizootic monitoring and clinical material has been demonstrated, allowing the developed method to be recommended for rapid diagnosis and epidemiological monitoring of WNF.

 

¹ URL: http://primerexplorer.jp/lampv5e/index.html

About the authors

Anna V. Mironova

Volgograd Plague Control Research Institute

Author for correspondence.
Email: mirnyuta@yandex.ru
ORCID iD: 0000-0002-6958-7861

researcher, Laboratory of gene diagnostics of particularly dangerous infections

Russian Federation, Volgograd

Olga S. Bondareva

Volgograd Plague Control Research Institute

Email: bondareva0s@mail.ru
ORCID iD: 0000-0001-5690-6686

Cand. Sci. (Med.), senior researcher, Laboratory of gene diagnostics of particularly dangerous infections

Russian Federation, Volgograd

Galina A. Tkachenko

Volgograd Plague Control Research Institute

Email: tkachenko_g@mail.ru
ORCID iD: 0000-0003-0199-3342

Cand. Sci. (Med.), Associate Professor, leading researcher, Department of biological and technological control

Russian Federation, Volgograd

Artem A. Baturin

Volgograd Plague Control Research Institute

Email: chemistry1987@mail.ru
ORCID iD: 0000-0001-9510-7246

Cand. Sci. (Biol.), senior researcher, Laboratory of gene diagnostics of particularly dangerous infections

Russian Federation, Volgograd

References

Supplementary files