A potent adaptive immune response requires the differentiation of B cells into Ig class-switched memory B cells bearing high-affinity antigen (Ag) receptors and plasma cells (PCs) secreting high-affinity antibody (Ab). The generation of these cells occurs in secondary lymphoid tissues within transient structures known as germinal centers (GCs). In addition to its role in normal humoral immunity, the GC has a critical role in lymphomagenesis, with the majority of B cell lymphomas thought to be GC or post-GC derived. As such, understanding how cellular signal-transduction pathways and genetic programs regulate GC B cell differentiation is of great importance not only to our understanding of adaptive immunity but also as a basis for understanding B cell lymphoma. Although our understanding of GC function has been greatly expanded through classic histological, flow cytometric and, more recently, advanced
in vivo imaging approaches, a detailed understanding of the molecular cues directing GC B cell fate can only be obtained through biochemical analyses of these cells
ex vivo1.
Although several cell-sorting methodologies are currently available, and mouse models for the induction and identification of GCs are well established, two main factors have contributed to the difficulty in GC B cell isolation: a relatively low frequency
in vivo and poor viability during and after sorting. Following immunization with a T cell–dependent Ag, such as a hapten-carrier with an adjuvant or heterologous erythrocytes, GC structures begin to form in as few as 3 d and continue to expand over the next several days as additional B cells enter the response and undergo significant bursts of proliferative expansion. Depending on the immunogen, with heterologous erythrocytes yielding the strongest response, the peak of the splenic GC response occurs 6–12 d after immunization
2,3. During this time, GC B cells account for approximately 5–15% of the B cell pool, which translates to 2–10% of splenic lymphocytes and, typically, to <1% of the total splenocytes
4,5. Although GC structures may persist for several weeks, the number of GC B cells decreases rapidly, nearly to preimmunization levels, within 1 week after the peak
5. Further limiting
ex vivo manipulation and interrogation of regulatory cascades in GC B cells is their poor survival after purification. Previous attempts at
ex vivo manipulation of GC cells have revealed that a majority of positively sorted mouse GC B cells die in culture within 4 h of isolation
6.
Development of the protocol
Several positive and negative cell-sorting methodologies have been used over the past several decades. These include panning (positive selection of target, or depletion of nontarget, cells based on binding to Ab- or lectin-coated polystyrene plates), complement-mediated lysis (removal of Ab-labeled nontarget cells by lysis mediated by purified complement proteins), fluorescence-activated cell sorting, commonly known as FACS (flow cytometry–based sorting of cells based on binding, or lack thereof, of fluorescently labeled Ab/Ag) and magnetic-activated cell sorting, commonly known as MACS (positive selection of target cells, or negative selection of target cells by depletion of nontarget cells, on the basis of binding to metal-containing beads and subsequent magnet-based removal). Identification of the GC B cell population requires flow cytometric and/or histological analysis that relies mainly on the identification of the induced surface markers GL7 and FAS and binding to peanut agglutinin, as well as on the lack of surface IgD
7–10. As a result, most of the previous methods for sorting these rare cells have relied on positive selection of cells or on a combination of depletion and positive selection
11–13. Inherent in any positive-selection approach is the possibility of altering the normal
in vivo signaling cascades and gene expression profiles (by ligation of surface Ags during sorting), which may cloud the interpretation of results. Therefore, a reliable negative-selection method is preferable to ensure accurate experimentation and interpretation of results. Several crucial factors affect the success of cell sorting, including purity, yield and maintenance of cell phenotype (including viability). With the intention of maximizing these factors, while using readily available reagents and equipment, we developed the procedure detailed herein and applied in our recent report
14. The method is a simple and fast yet reliable magnetic bead–based negative-selection scheme that allows the purification of untouched mature GC and non-GC B cells from the spleens of sheep red blood cell (SRBC)-immunized mice (). This protocol yields GC and non-GC B cells of 85–90% purity or greater. The sorting process can be carried out in ~1 h and yields a population of pure GC B cells in numbers large enough to allow
ex vivo manipulation, including biochemical analysis. Using this process, we were able to conduct the first biochemical (western blot) analysis of GC B cells, thereby leading to our discovery that the cAMP-PKA-GSK3 pathway is a critical regulator of cyclin D3 stability in GC B cells
14.
Application of the method
The sorting protocol detailed here is optimized for magnetic sorting of GC and non-GC B cells from the spleen of SRBC-immunized mice for the study of intracellular signal cascades ex vivo, using classical biochemical approaches following receptor-mediated stimulation and use of pharmacological inhibitors. Though this protocol utilizes the Miltenyi quadroMACS magnet and LS columns, it is likely to be adaptable to other magnetic technologies. In addition, the protocol should be adaptable to sorting GC B cells induced by a variety of other Ags from the spleen or lymph node. GC B cells purified using this protocol could also be used for additional experimentation including, but not limited to, adoptive transfer (transplantation) into syngeneic hosts, cell culture–based studies including migration and differentiation, gene expression profiling and possibly advanced proteomics.
Comparison with other methods
Panning The use of panning to purify lymphocytes was first developed more than three decades ago
15. Although this method has the potential to yield lymphocyte population of 98% purity, adaptation of this protocol to GC purification has proved less useful. Use of peanut agglutinin (PNA) binding to positively select human GC B cells resulted in just 50–80% purity even when a pre-enrichment step involving depletion of T cells and/or IgD
+ B cells (by panning or complement-mediated lysis) has been used to improve purification
11,12. In addition to poor purity, a drawback of this method is low yield, which results from inefficient elution of bound cells from the plate. Therefore, this method does not yield cells of adequate purity or number for subsequent biochemical analysis.
Complement-mediated lysis Complement-mediated lysis is an effective method for depletion of nontarget cells and has the ability to yield populations of cells exceeding 90% purity. It is also a good choice for depleting cells from multiple samples simultaneously or from large volume samples, as the only equipment required are a heated water bath and a centrifuge. However, the process of complement lysis involves incubation of cell samples for several hours at 37 °C. This is not favorable for GC B cells, which have poor viability ex vivo when not kept on ice or provided with extracellular stimuli.
FACS The first fluorescence-based sorting of GCs was conducted nearly 30 years ago when a single-color sort was performed to isolate cells based on relative binding of fluorescently labeled PNA
13. Despite being at the forefront of technology at the time, the experimental yield of just 1,000 GC B cells was certainly inadequate for extensive experimentation. FACS technology has made great advances since its advent in the 1970s, with the most advanced modern sorters capable of measuring 24 fluorescent parameters simultaneously, processing more than 200,000 cellular events per second, and sorting six populations exceeding 95% purity from the input sample (see
http://www.bdbiosciences.com/instruments/). Despite this precision and resultant purity, two major drawbacks still exist. The first is availability. Limited access to instrumentation, highly trained operators capable of sorting rare populations and the cost associated with this technology prevent its use by most investigators. The second crucial problem, sorting time, is problematic because of the nature of the GC B cell. As stated, GC B cells have poor viability and rapidly undergo apoptosis
ex vivo, including during sorting. As a result of the infrequency of GC B cells, long sort times are required to achieve high yield. Although the most advanced modern cell sorters boast sorting rates of 200,000 events per second, it is our experience that increased sorting speeds, especially with rare populations, necessarily result in either lower purity or increased abort rates, thereby resulting in lower yield, increased sort times and the need to use increased numbers of input cells. Therefore, sorting rates of <10,000 events per second are commonly used for sorting rare populations from relatively small starting samples. Despite the use of refrigerated chambers, aseptic collection and high sorting speeds, ~8 h of sorting time would be required to sort the 1 × 10
7 GC B cells (our average yield from eight SRBC-immunized mice) typically required for an inhibitor study, even when assuming an acquisition rate of 7,000 events per second and a GC B cell population accounting for 10% of the splenic B cells. The use of older and more commonly available sorters, which typically sort just 1,000 events per second for rare populations requiring high purity, in an experiment in which GC B cells account for 5% of the total B cells, would take nearly 100 h of sorting to obtain 1 × 10
7 GC B cells. In addition, sorting non-GC B cells from the same sample would either require the undesirable positive sorting of cells or would require sorting a second independently stained sample. This is in stark contrast to the speed at which GC B cells can be sorted using MACS technology. If a quadroMACS magnet with LS columns is used, ~5 × 10
6 GC B cells can be sorted every 30 min. Therefore, the decision to use FACS to purify GC B cells depends on the instrumentation available, the number of cells required and the level of purity desired. For these reasons, FACS-based isolation of GC B cells is most appropriate for gene expression profiling, whereas the MACS scheme we developed is better suited for
ex vivo manipulation and biochemical analysis, which require greater cell numbers.
Experimental design
Despite the simplicity of our approach, several considerations must be made during the sorting process. Several critical details including splenic disruption technique, specific Ab-labeling time and concentration and sample handling can have significant effects on the purity and yield of the cell sorting.
The protocol can be divided into five main steps: immunization (GC induction), isolation of tissue and preparation of cell suspension (cell preparation), labeling of cells with Ab-depletion cocktail (Ab labeling), sorting of labeled cells using MACS separator (magnetic sorting) and purity analysis.
GC induction This protocol was optimized using intraperitoneal injection of SRBCs in 129svj and outbred (129svj × C57BL/6) animals. We have found that a fresh preparation of 10% (vol/vol) washed SRBCs in PBS yields the best induction. The quantity of injection should be based on animal-use guidelines; typically 100 μl should be injected. Intravenous injection may induce a more robust splenic GC response and can also be used if desired. We have not evaluated the effect of adjuvants on this protocol. If the use of other immunogens, such as hapten-carrier or protein Ag, which require adjuvant, is desired, then the protocol may require additional optimization. We have not evaluated possible strain-specific variation in the protocol. Similarly, we have not evaluated the protocol for isolation of GC B cells from lymph nodes after subcutaneous or intramuscular injection.
Cell preparation The time of tissue harvest after immunization can have a substantial effect on yield and purity. We have found that harvesting 5–6 d after immunization with SRBC results in the best yield and purity. Tissues should be isolated immediately after euthanizing, placed in medium on ice and kept cold for the remainder of the procedure. Earlier isolation may result in poor yield of GC and poor purity of non-GC B cells, as a large fraction of IgD+ GL7− PNAhi early GC B cells may be present. Later isolation may also result in low yield or altered proportions of IgG class-switched cells in the preparation, which must be considered depending upon downstream application. As other Ags and adjuvants induce GC response of different kinetics, the harvest time will require optimization if other immunogens are used. We speculate that harvesting on day 8–10 post-immunization with hapten-carrier in alum would yield similar results as those described in this protocol, with the exception of decreased yield. In addition to the time of harvest, the method of tissue dissociation can markedly affect purity. We have found that physical dissociation of spleens between the frosted ends of glass slides yields the best purity. Although dissociation with mesh or filters and/or the use of dissociation agents such as collagenase or pronase may increase yield, we have observed reduced purity with these methods. It is likely that these preparation methods free additional unidentified cell types from the tissue that is not depleted by the Ab cocktail used in this protocol. Addition of Abs specific for these cell types could theoretically be used to improve purity should these cell preparation methods be used. Red blood cells must be removed before magnetic sorting or poor yield will result. RBC lysis should be carried out on ice. This protocol calls for ammonium chloride potassium lysis buffer (ACK) solution; however, commercially available RBC lysis solution may be used according to the manufacturer’s guidelines. Centrifugation and density-based removal of RBC (with Ficoll, for example) can be used, but we generally find that these methods reduce yield.
Ab labeling Cells should be kept on ice during staining. The use of biotinylated Abs against CD43, CD11c and IgD should yield a GC B cell population of ~90% purity. The use of biotinylated Abs against CD43, CD11c and GL7 should yield a non-GC B-cell population exceeding 90% purity. We use 90–95% of splenocytes for the GC sort and 5–10% of splenocytes for the non-GC sort, on the basis of the expectation that ~10% of B cells will be of the GC phenotype at the time of isolation. This can be altered depending on the immunogen used. The purity and yield are greatly affected by the Ab-labeling process. The use of incorrect concentration of the Ab in the staining cocktail or inadequate washing following staining may result in poor yield and/or purity. Inadequate washing or the use of too little Ab will result in poor purity, whereas use of too much Ab will result in poor yield. The concentrations stated in the procedure have been empirically determined. Although we utilize biotinylated Abs and anti-biotin microbeads according to the manufacturer’s protocols, we have found that the quantity of microbeads used can be reduced with minimal effect on purity and yield. We have used as little as 10 μl of microbeads per 107 splenocytes in 100-μl final volume, with comparable yield and only slight reductions in purity compared with the recommended use of 20 μl. In addition, other methodologies such as the use of fluorescently labeled Ab and anti-fluorophore beads may be used with further optimization. In other sorting trials, we have found the use of streptavidin beads to result in lower purity and yields compared with anti-biotin microbeads and have thus avoided their use here. If isolation of non-GC B cells from the same mouse is not required, the non-GC sorting can be omitted.
Magnetic sorting Although our protocol has been optimized using the Miltenyi midiMACS system with LS columns, we speculate that similar results should be attainable with magnetic depletion systems from other manufacturers. Our protocol is based on the manufacturer’s suggested protocol. We load a total of 1 × 108 splenocytes per column for GC sorting. This is the maximum number of bound cells suggested for LS columns and we assume that greater than 90% of loaded cells will be retained in the column. We have successfully loaded as many as 1.2 × 108 splenocytes for GC B cell sorting without loss of yield or purity. However, care must be taken to avoid overloading the columns. Exceeding the maximum binding capacity of the column will result in poor purity as a result of bead-labeled cells flowing through the column without being bound, and also poor yield, and the possibility of clogging the column is increased. To minimize loss of cells due to cell death, it is important that MACS buffer be prechilled to 4 °C and kept cold during sorting and also that cells be collected on ice. Cells should be used for downstream application immediately following sorting.
The publisher's final edited version of this article is available at 
![[filled triangle]](/corehtml/pmc/pmcents/utrif.gif)
CRITICAL STEP The SRBC suspension must be used immediately for immunization (Step 5).
