Efficient method for isolation of reticulocyte RNA from healthy individuals and hemolytic anaemia patients

Abstract Despite enormous progress and development of high‐throughput methods in genome‐wide mRNA analyses, data on the erythroid transcriptome are still limited, even though they could be useful in medical diagnostics and personalized therapy as well as in research on normal and pathological erythroid maturation. Although obtaining normal and pathological reticulocyte transcriptome profiles should contribute greatly to our understanding of the molecular bases of terminal erythroid differentiation as well as the mechanisms of the hematological diseases, a basic limitation of these studies is the difficulty of efficient reticulocyte RNA isolation from human peripheral blood. The restricted number of possible parallel experiments primarily concern healthy individuals with the lowest number of reticulocytes in the peripheral blood and a low RNA content. In the present study, an efficient method for reticulocyte RNA isolation from healthy individuals and hemolytic anaemia patients is presented. The procedure includes leukofiltration, Ficoll‐Paque gradient centrifugation, Percoll gradient centrifugation, and negative (CD45 and CD61) immunomagnetic separation. This relatively fast and simple four‐stage method was successfully applied to obtain a reticulocyte‐rich population from healthy subjects, which was used to efficiently isolate the high‐quality RNA essential for successful NGS‐based transcriptome analysis.


| INTRODUCTION
Reticulocytes are non-nucleated immature red blood cells in peripheral blood of mammalian species. Young cells contain residual RNA and micro-organelles such as ribosomes, lysosomes, and mitochondria. The residual RNA is not only functional, but also necessary for the final stages of reticulocyte maturation. 1,2 Transformation from a reticulocyte to a fully mature red cell is a complex process requiring substantial remodelling of the erythrocytic cytoplasm and membrane. 3 Maturation of the reticulocyte is a 2-3-day-long process, during which hemoglobin synthesis occurs, and during the last days old reticulocytes eliminate their residual RNA and are released into the circulation, to become mature red blood cells. 4 The presence of an intracellular network of residual reticulocyte RNA allows one to differentiate between various stages of the development of reticulocytes. The detection of reticulocyte RNA with brilliant cresyl blue allows one to distinguish reticulocytes from mature red blood cells as well as enabling the identification of the youngest, highly fluorescent reticulocytes (HFR), supplied early from bone marrow in conditions of increased erythropoietic stimulation, eg, hemolytic anaemia.
An example of such conditions is increased hemolysis due to red blood cell pathology. Importantly, in hemolytic anaemia both HFR and the reticulocyte count are markedly increased and inversely correlated with hemoglobin (Hb) level. 5 Hence, reticulocytes (as well as their transcriptome) may be useful in the diagnostics of hematopoietic system diseases including hemolytic anaemia.
Adult peripheral blood is characterized by a low reticulocyte count resulting in an extremely low yield of reticulocyte RNA isolation as compared to cord blood. 6 Therefore, few parallel experiments on healthy individuals' reticulocytes are possible due to having the lowest number of reticulocytes in their peripheral blood.
Furthermore, although reticulocyte transcriptome analyses could greatly improve our understanding of reticulocytes' maturation, they require relatively high quality of reticulocyte RNA free of RNA derived from other blood cells. Only few procedures for peripheral blood reticulocyte isolation are available in the literature. 7

| RNA isolation and cDNA synthesis
Total RNA was extracted from the reticulocyte-rich suspension and collected using the miRNeasy Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer's recommendation. In the first isolation step,~0.4 mL of the reticulocyte-rich pellet with the supernatant was divided into two aliquots. In further steps we followed the manufacturer's procedure until the material was applied onto an RNeasy Mini Spin Column. During this step, the materials from two aliquots were pooled (to enhance yields of RNA isolation), to ensure fulfilling the minimal column capacity required for efficient RNA isolation. RNA concentrations were calculated based on the absorbance at 260 nm. RNA samples were stored at −70°C. The first strand of cDNA was synthesized using a RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, Waltham, MA, USA).

| Assessment of purity of isolated reticulocyte population by RT-PCR
To verify the purity of the reticulocyte population the RNA isolates from reticulocyte populations was tested using specific primers for marker gene transcripts: HBB, PPARA, PTPRC, ITGB3, and ACTB. Primer sequences are listed in Supporting Information Table S2. A twostep RT-PCR was applied (DreamTaq DNA Polymerase; Thermo Fisher Scientific).

| Next generation RNA sequencing analyses
Next generation sequencing was performed by Heflin Center for Genomic Science Core Laboratories, University of Alabama at Birmingham, AL, USA. Directly prior to the RNA-Seq procedure, the total F I G U R E 1 Scheme of four-step purification procedure used to obtain reticulocyte-rich suspension from human peripheral blood SKULSKI ET AL. | 489 RNA integrity was verified by on-chip electrophoresis (Bioanalyzer, Agilent Technologies, Santa Clara, CA, USA) and only samples with high RNA integrity (RIN) values were further processed (>7.5). In addition, quality score analysis was performed confirming the high quality of RNA samples. Following rRNA depletion, the remaining RNA fraction was used for library construction and subjected to 100 bp paired-end sequencing on an Illumina HiSeq 2000 instrument.
Sequencing reads were aligned to the human reference genome assembly (hg19) using TopHat. 12 Transcript assembly and estimation of the relative abundances were carried out with Cufflinks. 13 15 Briefly, all genes in each SuperPath are given a similar weight in the analysis, and the matching score is based on the cumulative binomial distribution, which is used to test the null hypothesis that the queried genes are not over-represented within any Super-Path pathway. Unification is employed on all of the sources found in GeneCards. When using GeneAnalytics we focus our analysis only on expression signatures that were assigned with high scores, which reflects high biological pathway/phenotype coverage. The gene nomenclature was adjusted to the webserver requirements (GeneCards).

| Reticulocyte isolation
Fresh peripheral blood was collected from healthy volunteers, patients with hereditary spherocytosis and hemolytic anaemia. The mean age of the volunteers/patients (6 men) was 34 years (25-40).
Differences in the reticulocyte count in hemolytic anaemia patients and healthy individuals are shown in Figure 2A. To obtain a reticulocyte-rich population free of nucleated blood cells we applied a fourstep procedure (Figure 1, Section 2). In the first step, the whole blood was filtered through a leukocyte removal filter. Then the resulting blood filtrate was centrifuged on the Ficoll-Paque Premium density gradient followed by further purification of reticulocytes by Percoll density gradient centrifugation. 16 The former step was repeated once to remove most of the red blood cells, as their presence prevented correct execution of the next steps. The above-mentioned steps of the presented procedure were still not sufficient to obtain a pure reticulocyte population. Monocytes and platelets (which also contained RNA) were still present in our microscopic preparations. Hence, to remove the residual nucleated cells and platelets the MACS magnetic cell separation technology was applied.
Initially, a method of positive selection of reticulocytes with the CD71 MicroBeads was used. However, this approach was successful for patients with high reticulocytosis only. In healthy subjects, CD71 + cells were so scarce that no effective RNA isolation was possible. For this reason, we decided to use negative selection of leukocytes and platelets using CD61 and CD45 MicroBeads. The resulting reticulocyte preparation, although containing a reasonable number of red blood cells (Supporting Information Figure S1), provided a starting material for efficient RNA isolation in which a highly purified reticulocyte population is needed. An effective method to differentiate various stages of the development of reticulocytes is the use of Thiazole Orange (TO) staining, 17 which can be substituted for another commonly used reticulocyte classification system known as the reticulocyte maturity index (RMI). 18 The most effective method of determining the cord reticulocyte subset with minimal artifacts is the use of anti-CD71 antibodies and magnetic separation. 6 The CD71 positive group characterizes the early human reticulocytes (Heilmeyer groups I to III) and the mature CD71 negative subset (Heilmeyer group IV) which were found in normal circulation. 19 To analyse the maturity status and the number of reticulocytes, the sample was incubated with Thiazole Orange (TO) and CD71-PE markers and counted by flow cytometry (Figure 2B-D). Results for each HA patient (N61 and N62) were compared independently to both controls (C1 and C14).

| RNA isolation and cDNA synthesis
In order to test the usefulness of the described procedure of reticulocyte purification the functional quality of RNA was tested. RNA isolation was carried out as described in the Section 2. The obtained RNA quantities were sufficient for RNA-Seq analysis for all study groups ( Figure 3). The amount of RNA obtained from patient C14 diagnosed with HS was relatively low despite the increased reticulocytosis, which in our opinion is a consequence of decreased osmotic resistance of the red blood cells/reticulocytes resulting in excessive hemolysis observed during the reticulocyte isolation procedure.

| Assessment of purity of isolated reticulocyte population by RT-PCR
The reliability of RNA-Seq analysis is based on obtaining a reticulocyte population free of contamination with other blood cells. Hence, to assess the reticulocyte purity we tested the obtained RNA isolates for the presence of high-and low-abundant transcripts characteristic for these blood cells as well as for the absence of transcripts specific for platelets and leukocytes. As shown in Figure 4 and Supporting Information Figure S2, the RNA obtained from the isolated reticulocyte population contained only the gene transcripts that were reticulocyte-specific: "high abundance" HBB, and "low abundance" PPARA (peroxisome proliferator-activated receptor alpha, by an example which was taken from the end of the reticulocyte cDNA library; NCBI, Library 11923). As a quantitative control β-actin (ACTB) primers were used. Furthermore, to verify efficient removal of leukocytes and platelets the RNA isolates from reticulocyte populations were tested. Transcripts specific markers for platelets (integrin β3 (ITGB3)) and for leukocytes (CD45 (PTPRC)) were absent.

Expression status for erythroid and non-erythroid transcripts
Reticulocyte transcriptome analysis using high-efficiency gene expression assays was performed by RNA-Seq for four male subjects F I G U R E 3 RNA amounts for each examined group. Healthy individuals (n = 4), C14-hereditary spherocytosis patient and HAhemolytic anaemia patients (n = 2) F I G U R E 2 Reticulocyte characteristics. A, Examples of reticulocyte quantification in human peripheral blood. Examination was performed for healthy individuals (C1-C3) and for hemolytic anaemia patients (N61 & N62). The number of reticulocytes was determined from the dot plots: FSC (forward scatter) vs TO fluorescence. B, Analysis of CD71 surface exposure in WBC and platelet-free reticulocyte preparations obtained from the healthy individuals and HA patients allows determination of reticulocyte number and maturity status. Dot plots showing CD71 surface expression and RNA content (TO) in reticulocytes isolated from healthy individuals and HA patients (N61, N62). These data indicate a significantly higher number of immature (early) reticulocytes in peripheral blood of HA patients than in healthy individuals. C, Quantification of purified reticulocyte measured by Thiazole Orange (TO, RNA content) fluorescence reveals 3.6 and 6.7 times higher signals for N61 and N62 HA patients, respectively, than in healthy individuals (controls: n = 3). D, CD71 surface expression in purified reticulocytes for N61 and N62 HA patients is significant SKULSKI ET AL.  (Table 1). These results indicate that the isolated population of reticulocytes is characterized by high homogeneity.
Using the GeneAnalytics web server, the results of RNA-Seq analysis were convincingly (high score) assigned to 111 metabolic pathways, 104 biological processes, 31 molecular function pathways, 75 diseases, and 22 phenotypes (Supporting Information Table S3).
Importantly, our set of gene transcripts was associated with 22 phenotypes that correlate with erythroid cells.

F I G U R E 4
Example of reticulocyte cDNA quality control. Agarose gel electrophoresis of RT-PCR products (image inverted, black/white) obtained using reticulocyte cDNA from patients (HA patient, N62, and HS patient, C14) and controls (healthy individuals: C1 and C2) and primers encoding sequences of the following genes: CD45 (PTPRC gene; 217 bp)-leukocyte marker; β-globin (HBB gene; 397 bp)-erythroid marker; integrin-β3 (ITGB3 gene; 198 bp)-platelet marker. As the "low abundance" gene transcript PPARA (peroxisome proliferator activated receptor alpha; 324 bp) was chosen and was taken from the end of the reticulocyte cDNA library (NCBI, Library 11923). For the loading control β-actin primers (ACTB gene; 479 bp) were used. All primer sequences used in this experiment are included in Supporting Information Table S2. As a standard "GeneRuler 100 bp DNA Ladder" (Thermo Fisher Scientific, Waltham, MA, USA) was used. Results show no leukocyte or thrombocyte RNA contamination for all analysed cases. The lowest band corresponds to primers MicroBeads/CD61 MicroBeads.
3. Finally, variability in the reticulocyte population between healthy individuals and patients with hematologic diseases should be considered. Purification of reticulocytes from patients with hemolytic anaemia without changed osmotic resistance of red blood cells is relatively simple due to the increased reticulocyte count and no excessive hemolysis during isolation. Patients with hemolytic anaemia associated with increased osmotic fragility including conditions such as hereditary spherocytosis or hereditary elliptocytosis are characterized by high reticulocytosis, but usually during the isolation procedure increased hemolysis occurs, which significantly reduces the efficiency of isolation. In our procedure, we tried to optimize the method, to minimize hemolysis. However, the most challenging aspect is achieving efficient separation of the reticulocytes from the peripheral blood of individuals with a normal reticulocyte count, which has been pointed out in some reports. 6   and CD71 ++++ library (Lib.8975) ( Figure 5A).  Importantly, the extended list of identified transcripts obtained in our study provided much better coverage of gene ontology terms, as reflected by higher scores. Furthermore, using the GeneAnalytics web server we identified in our transcript set of genes associated with 22 phenotypes, whereas microarray data analysis resulted in identification of transcripts associated with 19 phenotypes (Table 2).

As shown in Supporting Information
Importantly, the genes identified by RNA-Seq were associated specifically with erythroid phenotypes.
Taken together, gene ontology analysis supports the conclusion that the experimental approach proposed in this study provides an improved insight into biological processes that govern reticulocyte maturation, function and pathology. Undeniably, obtaining depth and reliable insight into related transcriptome profiles should significantly help to understand these mechanisms.
In conclusion, we have presented an efficient procedure of reticulocyte isolation from peripheral blood. This procedure should be useful in particular for normal, healthy individuals, whose blood is poor in reticulocytes, and for hemolytic anaemia patients including HS patients with increased hemolysis. This will increase the statisti-

CONFLI CT OF INTEREST
The authors declare that they have no conflict of interests concerning the contents of this article.