Characterisation of the Immunomodulatory Effects of Meningococcal Opa Proteins on Human Peripheral Blood Mononuclear Cells and CD4+ T Cells

Opa proteins are major surface-expressed proteins located in the Neisseria meningitidis outer membrane, and are potential meningococcal vaccine candidates. Although Opa proteins elicit high levels of bactericidal antibodies following immunisation in mice, progress towards human clinical trials has been delayed due to previous findings that Opa inhibits T cell proliferation in some in vitro assays. However, results from previous studies are conflicting, with different Opa preparations and culture conditions being used. We investigated the effects of various Opa+ and Opa- antigens from N. meningitidis strain H44/76 in a range of in vitro conditions using peripheral blood mononuclear cells (PBMCs) and purified CD4+ T cells, measuring T cell proliferation by CFSE dilution using flow cytometry. Wild type recombinant and liposomal Opa proteins inhibited CD4+ T cell proliferation after stimulation with IL-2, anti-CD3 and anti-CD28, and these effects were reduced by mutation of the CEACAM1-binding region of Opa. These effects were not observed in culture with ex vivo PBMCs. Opa+ and Opa- OMVs did not consistently exert a stimulatory or inhibitory effect across different culture conditions. These data do not support a hypothesis that Opa proteins would be inhibitory to T cells if given as a vaccine component, and T cell immune responses to OMV vaccines are unlikely to be significantly affected by the presence of Opa proteins.


Introduction
Neisseria meningitidis causes approximately 500,000 cases of meningitis and septicaemia worldwide annually, with a case-fatality rate of approximately 10% [1]. Most disease is caused by capsular group A, B, C, W, X and Y organisms. Protein-polysaccharide conjugate vaccines are in routine use globally for capsular groups A, C, W and Y, and group B is the major cause of disease in most temperate countries [2][3][4][5][6]. The Opacity-associated (Opa) adhesin proteins are major phase-variable proteins found in the outer membrane of N. meningitidis. Each bacterium contains four opa genes (opaA, opaB, opaD and opaJ), which may encode identical or different Opa proteins, and can therefore express up to four different Opa variants at any one time [7][8][9]. Opa proteins are critical in meningococcal pathogenesis, mediating bacterial adherence to the nasopharynx and modulating human cellular immunity via interactions with T cells and neutrophils [10,11]. They are also important during infection with the related species N. gonorrhoeae, mediating persistence of bacteria in the genitourinary tract via the avoidance of an effective immune response [12].
N. meningitidis can persist in the human nasopharynx without causing symptoms for several months, and N. gonorrhoeae can cause prolonged mucosal infection of the genito-urinary tract. This ability to persist relies on their adaptability to the host and their capacity to evade the immune system. Carcinoembryonic antigen-related cell adhesion molecules (CEACAMs) are cell surface glycoproteins found on a range of cell types. Binding of these proteins by various ligands can result in up-or down-regulation of intracellular signalling pathways [12]. Opa protein binding to CEACAMs on the surface of host cells confers the ability to associate with human epithelial, endothelial and leucocytic cells encountered during neisserial infection, indicating a direct effect on the immune response [13]. Although Opa proteins are able to bind to a number of different CEACAMs, CEACAM1 has a broad expression distribution in normal tissues and is the only member of the family present on the surface of T cells. The response of T cells, and particularly CD4 + T cells, is important during infection with pathogenic Neisseria as these cells are involved in directing the magnitude and quality of humoral immune response. Antibodies directed against surface structures of N. meningitidis are important in immunity but gonococci do not induce a strong, protective antibody response following infection [14]. T cells are also important in the generation of immunological memory and possibly cell-mediated immunity, which is therefore relevant to vaccine development [15]. The interaction between meningococci and human T cells and the particular role of Opa proteins in this interaction has therefore been the subject of intense, and conflicting, study in the last decades [16][17][18][19][20][21][22][23][24]. Furthermore, Opa proteins have been suggested as potential meningococcal vaccine candidates as they elicit high levels of bactericidal antibodies in mice [13]. However, sequence variability of some of the surface-exposed loops and uncertainty regarding their immunomodulatory effect on human T cells has delayed further development into clinical trials.
In this study we investigated the effects of recombinant and liposomal Opa proteins, in addition to Opa+ and Opa-outer membrane vesicles (OMVs) and bacteria based on isogenic strains, on the immunomodulatory interaction between N. meningitidis and human peripheral blood mononuclear cells (PBMCs) and CD4 + T cells. In an attempt to clarify the effects of Opa proteins on CD4 + T cells, a number of assays were undertaken using different cell culture conditions, and a variety of Opa+ and Opa-antigens.

Study subjects
Written informed consent was obtained from 46 healthy adult volunteers recruited to the study (aged 18 to 66 years) prior to collection of a single blood sample. Anyone with a history of previous IMD, a known immunodeficiency, or who was enrolled in another study which may affect their immune responses was excluded. The study was approved by the Oxfordshire C Research Ethics Committee (REC No: 07/H0606/84; UKCRN ID 4609).

Isolation of peripheral blood mononuclear cells and purification of CD4 + T cells
A maximum of 40 ml of blood was collected from each study participant, and heparinised blood (1000 units/ml heparin) was diluted in an equal volume of culture medium buffer (RPMI-1640 medium, HEPES modification, 25 mM HEPES, 50 units/ml penicillin, 50 μg/ml streptomycin, 2 mM L-glutamine [Sigma-Aldrich, Gillingham, UK]). PBMCs were isolated by density gradient centrifugation (Lymphoprep, Axis-Shield, Dundee, UK). Cells were subsequently either labelled with carboxyfluorescein succinimidyl ester (CFSE) prior to culture and stimulation with antigen, or directly processed for purification of CD4 + T cells. CD4 + T cells were purified from PBMCs by negative selection according to manufacturer's instructions (CD4 + T cell isolation kit II, Miltenyi Biotec, Woking, UK). Following separation, cells were labelled with CFSE prior to culture and stimulation with antigen.
Cell labelling with carboxyfluorescein succinimidyl ester (CFSE) To assess proliferation by flow cytometry, fluorescent intracellular labelling of PBMCs and purified CD4 + T cells with CFSE was performed prior to culture and stimulation with antigen [25]. In brief, the cells were resuspended in CFSE buffer (PBS containing 2mM EDTA and 0.5% newborn calf serum [NBCS]), and diluted to a concentration of 1 x 10 6 cells/ml in CFSE buffer containing 2 μM CFSE. Cells were incubated at 37°C with 5% CO 2 for 10 minutes and washed twice in culture complete medium (containing 10% NBCS). Cell suspensions (2 x 10 6 cells/ml for PBMCs and 1 x 10 6 cells/ml for purified CD4 + T cells) were prepared with or without IL-2 (final concentration of 1,000 units/ml) (Roche, Welwyn Garden City, UK).

Cell culture and T cell stimulation
Cells were cultured in flat-bottomed 96-well tissue culture plates (Thermo Scientific Nunc, Loughborough, UK) at a cell density of 2 x 10 5 or 1 x 10 5 cells per well for PBMC and CD4 + T cells, respectively, in complete culture medium (100 μl per well). To induce the expression of CEACAM1, cells were pre-incubated with IL-2 (1,000 units/ml) for 48 hours at 37°C with 5% CO 2 prior to antigen addition. For ex vivo studies, antigens were diluted prior to addition to the culture plate with or without T cell stimulation using purified anti-human CD3 (UCHT1) and anti-human CD28 (CD28.2) antibodies for a further 96 hours (Table 1). Per participant, each antigen and assay controls were tested in triplicate. In order to comprehensively investigate the effects of Opa proteins on CD4 + T cell proliferation, different Opa+ or Opa-antigens were tested using a combination of culture conditions. The following culture conditions were tested: (a) ex vivo culture of PBMCs with antigen for 5 days; (b) PBMCs cultured for 4 days with antigen and IL-2 (following 48 hours pre-incubation with IL-2); (c) PBMCs cultured for 4 days with antigen, IL-2, anti-CD3 and anti-CD28 (following 48 hours pre-incubation with IL-2); (d) purified CD4 + T cells cultured for 4 days with antigen, IL-2, anti-CD3 and anti-CD28, (following 48 hours pre-incubation with IL-2). Priority was given to assays involving ex vivo PBMCs, as the convention used to assess T cell stimulation and of most interest to vaccine development. Co-ligation culture conditions including anti-CD3 and anti-CD28 were performed as a priority over those with IL-2 alone when the number of cells recovered from blood was insufficient for all assays.

Antigens for T cell stimulation
An extensive antigen panel consisting of purified recombinant Opa proteins, liposomes, OMVs and killed bacteria from isogenic strains constructed from parent strain H44/76 were studied ( Table 1). The isogenic strains used in this study included the wild-type parent strain H44/76, an Opa-strain (M014), a strain expressing OpaD only (M002), and a strain expressing both OpaA and OpaD (M001) [26]. opa genes associated with hyperinvasive serogroup B meningococci were cloned, expressed in Escherichia coli BL21 (DE3), purified from inclusion bodies using affinity chromatography [13] and incorporated into liposomes [27]. Non-CEACAM binding mutated forms of OpaA and OpaD were constructed by site directed mutagenesis, replacing Gly-172, Ile-174 and Gln-176 with alanine, as previously described [28]. Phytohaemagglutinin (PHA) [29,30] and Staphylococcal enterotoxin B (SEB) were used as positive controls for the assays. A rabbit polyclonal anti-CEACAM antibody (DAKO, Ely, UK) was used as an inhibitory control for T cell responses while an isotype matched antibody control was used as the comparator [31].

Analysis of cells using flow cytometry
Generations of proliferating CD4 + T cells were assessed by flow cytometry [25]. CD4 + T cells were positively identified by staining with a mouse anti-human CD4-APC conjugate (BD Pharmingen, Oxford UK) and dead cells were excluded from the analysis using 7-aminoactinomycin D (7-AAD) (Sigma-Aldrich, UK). Flow cytometry was performed using the FACSCalibur (Becton Dickinson, UK). Data from 10,000 CD4 high 7-AAD low events were collected for each sample. Data were collected and stored using CellQuestPro for Macintosh (Becton Dickinson). Assay controls for each participant included unstained (minimum auto-fluorescence) and CFSE-stained unstimulated controls (media only control). For analysis, a histogram plot was created and a marker was used to identify the fluorescence peak of the media only control (maximum fluorescence of the undivided cells). The average number of proliferating events in the final culture population for triplicate measurements was determined. Cell proliferation is defined as a percentage of the total number of CD4 + T cells analysed.

Bactericidal antibody responses following immunisation of mice
Groups of ten 6-7 week old female BALB/c mice (Charles River, Margate, UK) were immunised subcutaneously with 0.2ml of recombinant Opa protein on days 0, 21 and 35 (5 μg per dose). The adjuvant used was the oil-in-water emulsion Sigma Adjuvant System (Sigma-Aldrich), which was prepared according to the manufacturer's instructions but in a total volume of 2.4ml of antigen solution per vial, and was reconstituted with the antigen on the day of immunisation. Blood was collected by cardiac puncture on day 42 and serum separated by centrifugation at 16,100 x g for 10 minutes. Bactericidal activity in mouse serum pooled within each immunisation group was quantified by SBA [32]. Briefly, pooled murine sera was heated at 56°C for 30 minutes to deactivate endogenous complement and then diluted to give a range from 1:4 to 1:1024. Diluted sera were incubated with mid-log phase meningococci (125 cfu) and baby rabbit complement (lot number 11330, PelFreez Biologicals, Rogers, AR) at a final concentration of 25% (v/v). Reactions were incubated for 60 minutes at 37°C in a humidified 5% CO 2 atmosphere. A sample of each reaction was spread by tilting onto BHI agar plates, which were incubated overnight. The bactericidal antibody titre was reported as the reciprocal of the highest serum dilution at which 50% bacterial survival was observed. Each serum sample was analysed in duplicate against each target strain.

Statistical analysis
CD4 + T cell proliferation was calculated for each antigen as the mean of the triplicate wells, by subtracting the mean proliferation of control wells containing media only. Opa+ antigens were compared to Opa-antigens in the same way. For wild-type and mutant forms of recombinant or liposomal Opa proteins, the mutated forms were considered Opa-in these analyses. Statistical significance was calculated using a paired two-tailed Student's t-test to compare cells from the same donor, with a p-value of <0.01 considered as statistically significant. Analysis was carried out using R (version 3.1.2).

PBMCs ex vivo CD4 + T cell proliferation
Proliferative responses to all antigens were low in these conditions, with a maximum amplitude of mean stimulation/inhibition of 1.8%, with the exception of PHA ( Table 2, Fig 1). There was statistically significant inhibition of T cell proliferation by the liposomal wild-type OpaD, although the magnitude of this inhibition was low (-1.6%), compared to the stimulatory effect seen with the PHA control (+61.3%). Otherwise there was no significant stimulation or inhibition by any of the Opa+ or Opa-antigens. When Opa+ antigens were compared to their Opacounterparts there were no significant differences, with the exception that Opa+ bacteria were relatively stimulatory. The magnitudes of these differences were small, +0.5% and +1.3% for OpaA+ OpaD+ and OpaD+ strains respectively. Although culture of PBMCs ex vivo without IL-2 and/or stimulation of the T cell receptor (e.g. with PHA or anti-CD3) is a standard method for assessing antigen-specific responses following a vaccination, proliferative responses measured in vitro in the absence of such a recent stimulus can be more difficult to detect [33]. Further experiments were therefore performed with additional IL-2 to amplify proliferative responses [34].

PBMCs with IL-2 only
A limited number of samples were tested with IL-2 only (up to 5 participants per antigen condition due to limited blood volumes available), to amplify responses by increasing activation and proliferation of T cells. In these conditions, all recombinant proteins caused inhibition of CD4 + T cell proliferation compared to media alone (between -16.3% and -6.5%), in addition to the liposomal wild-type OpaD (-18.0%) ( Table 3, Fig 2). Wild-type recombinant Opa proteins were significantly more inhibitory than the mutated equivalents, with a difference of approximately 9% for OpaA and OpaD. Wild-type liposomal OpaD was also more inhibitory than the mutated form (difference 14.9%). None of the OMVs or inactivated bacteria were significantly stimulatory or inhibitory compared to the media only control, but Opa+ OMVs were slightly more stimulatory than Opa-OMVs. In contrast OpaA+ OpaD+ bacteria were less stimulatory than Opa-bacteria, although this difference was not observed with OpaD+ bacteria. All differences between Opa+ and Opa-OMVs and bacteria were relatively small in magnitude (between 3.0% and 5.6%).

PBMCs and purified CD4 + T cells with IL-2, anti-CD3 and anti-CD28
In these culture conditions, CD4 + cells are highly stimulated non-specifically, so inhibitory effects would expect to be seen. When purified CD4 + T cells were used, there was significant inhibition of CD4 + proliferation by all recombinant proteins and liposomes, with the exception of mutated rOpaA (although the trend was also towards inhibition in this case). Inhibition of proliferation ranged between -9.6% (rOpaD mutant) and -46.1% (lOpaD wild-type) (Tables 4  and 5, Figs 3 and 4). The Opa+ recombinant and liposomal proteins were significantly more inhibitory than the Opa-forms (with the exception of liposomal OpaA which also showed the same trend), with differences of between 5.6% and 30.1%. In addition, there was inhibition of CD4 + T cell proliferation by all OMVs (reaching statistical significance for all except wildtype), but Opa+ OMVs were less inhibitory than Opa-OMVs. All of the bacterial preparations tested were able to provide additional stimulation to CD4 + T cells in these conditions, with no significant differences between Opa+ and Opa-bacteria. When mixed PBMCs were used in the same highly stimulated culture conditions, results for the recombinant and liposomal proteins were similar to those found using purified CD4 + T cells, with inhibition observed for all preparations, although fewer achieved statistical significance. On comparison of the mutated and wild-type proteins there was only a significant difference for liposomal OpaD, with the wild-type protein again being more inhibitory than the mutated protein (difference 35.6%). All of the bacterial strains tested were able to provide additional stimulation to CD4 + T cells, as observed with purified CD4 + T cells, and wild-type and OpaD+ bacteria were slightly less stimulatory than Opa-bacteria. Most OMVs in these conditions were not significantly stimulatory or inhibitory, although a slight stimulatory effect was observed with Opa-OMVs. A comparison of results from this study and previous studies of Opa-mediated effects on CD4 + T cells is summarised in Table 6.

Bactericidal antibody responses following immunisation of mice with recombinant Opa proteins
The immunogenicity of recombinant forms of wild-type and mutated OpaA and OpaD from H44/76 were tested using Opa+ and Opa-strains as target strains in the serum bactericidal antibody (SBA) assay. Data using wild type proteins has been reported previously, and are included here for comparison to the mutated forms [26]. The recombinant proteins resulted in titres of 1:256 when the target strain expressed the same Opa variant used for immunisation (Table 7), and <1:4 otherwise. Introduction of mutations in the hypervariable HV2 regions led to complete abrogation of SBA activity for OpaA, but had no effect on the SBA titre for OpaD against either OpaD+ target strain.

Discussion
In this study, a range of Opa+ and Opa-antigens was used in culture with both PBMCs and purified CD4 + T cells to demonstrate that recombinant and, to a lesser extent, liposomal OpaA and OpaD proteins from N. meningitidis strain H44/76 inhibited CD4 + T cells in vitro when cells were pre-stimulated with IL-2, with or without anti-CD3 and anti-CD28 antibodies. When the HV2 region of these Opa proteins was mutated to abrogate CEACAM1 binding, these inhibitory effects were usually reduced, although inhibition still occurred. These same mutations had variable effects on immunogenicity of these proteins when used for immunisation in mice-bactericidal antibodies were still induced with OpaD, but not for OpaA. It is not possible to directly contrast the T cell studies with mouse immunisations because of the different CEACAM repertoire found in humans and mice. In contrast, OMVs and inactivated bacteria based on Opa+ and Opa-strains did not have the same inhibitory effects on CD4 + T cells. All bacterial strains were typically stimulatory and there were no consistent differences between Opa+ and Opa-bacteria or OMVs. It is likely, therefore, that other proteins contained within these preparations are immunodominant with respect to these in vitro interactions. This study was the first to study the culture of freshly isolated PBMCs ex vivo to assess Opamediated interactions, and under these conditions none of the Opa+ or Opa-antigens exerted a major stimulatory or inhibitory effect on CD4 + T cells. This is a standard method of assessing in vitro T cell responses to vaccine candidates and the assessment of antigen-specific proliferation [35] following vaccination. In this study, T cell responses were assessed in healthy adults without recent meningococcal vaccination or infection, which may explain the relative lack of proliferative responses observed in the PBMC ex vivo conditions. This study, therefore, does not support the argument that Opa proteins would have an inhibitory effect in vivo if given as a vaccine to healthy adults. Immune responses to previous OMV vaccines have included the induction of anti-Opa antibodies, which also suggests that Opa does not have a significant immunosuppressive effect when included within OMVs.
Previously published data regarding the effects of meningococcal Opa proteins on CD4 + T cell proliferation have been conflicting, creating a significant hurdle in the development of any Opa-containing vaccines ( Table 6). Studies comparing the meningococcal proteins PorA, PorB, Opa and Opc showed that Opa induced the strongest T cell proliferative responses [16,17] and whole bacteria and OMVs have been shown to stimulate CD4 + T cell proliferation, independent of their Opa phenotype [18]. Several Opa-containing OMV vaccines are Table 4. CD4 + T cell proliferation following incubation of PBMCs with different antigens, plus IL-2, anti-CD3 and anti-CD28.

Antigen
Mean % CD4 + T-cell proliferation immunogenic in humans [19][20][21][22][23]. In other studies, when CD4 + T cells were exposed to OMVs from Opa-expressing meningococci or to gonococci expressing CEACAM-binding Opa variants, their activation and proliferation in response to a variety of stimuli were effectively halted [24,31]. There may be a number of reasons to explain the differences between various studies. Different studies have used different Opa variants and Opa proteins from H44/76 have not previously been shown to have an inhibitory effect. The proposed mechanism of inhibition is via an Opa-CEACAM1 interaction, and both Opa proteins used in this study bind to CEACAM1 [28]. When OMVs and bacteria are used, there may be variation of Opa expression levels relative to other outer membrane proteins, resulting in different effects. Further major differences between studies are the culture conditions and methods of assessment of proliferation. PBMCs and purified CD4 + T cells have been used, with most studies including assays with IL-2 preincubation and/or addition of anti-CD3 and anti-CD28. Differences in duration of culture may Table 5. CD4 + T cell proliferation following incubation of purified CD4 + T cells with different antigens, plus IL-2, anti-CD3 and anti-CD28.

Antigen
Mean % CD4 + T-cell proliferation also affect CEACAM1 expression on the T cell surface. While these conditions are useful in exploring specific interactions, they are unlikely to represent what occurs in vivo, and should be interpreted with caution when predicting the potential effects of Opa proteins during meningococcal infection or following immunisation in humans. IL-2 is a T cell growth factor and increases CEACAM1 expression, potentially enhancing any CEACAM1-mediated Opainduced effects [31], supported by the inhibition of proliferation observed with the anti-CEA-CAM antibody in these conditions. IL-2 is produced in vivo during T cell responses, and it is likely that the PBMC ex vivo assay represents more physiological levels of IL-2, compared to those assays when exogenous IL-2 was added. Anti-CD3 and anti-CD28 antibodies provide signals for T cell activation, as well as further increasing CEACAM1 expression on the T cell surface [31]. Incubation of cells with IL-2, anti-CD3 and anti-CD28 therefore tends to lead to inhibitory rather than stimulatory effects. The greater baseline stimulation of cells where anti-CD3 and anti-CD28 have been added is also the most likely explanation for the greater amplitude of inhibition observed in these conditions, compared to with IL-2 alone. The addition of PHA in the presence of anti-CD3 and anti-CD28 resulted in less additional proliferation than when they were absent, confirming these cells were already highly stimulated. Methods of measuring CD4 + T cell proliferation have varied between each of the published studies. A flow cytometry approach based on CFSE dilution was adopted in this study to ensure that dead cells were excluded from the analysis, but this is more difficult to achieve when using other methods described. Our data support the conclusion that isolated Opa proteins have a different effect to Opa expressed on the bacterial surface with other proteins. Considering that Neisseria have the capacity to randomly and reversibly switch many surface antigens 'on' and 'off' by a process of PV, it is also possible that the differential expression of uncharacterised factors other than Opa may influence T cell responses during infection. However, we used OMVs and bacteria from isogenic strains, constructed from the same parent strain, to reduce this risk. We have previously shown that expression of the major proteins in these strains is the same [26]. The results from these experiments do not support a proposed hypothesis that Opa proteins would inhibit CD4 + T cell proliferation when part of whole bacteria (as would be the case during infection) or contained within an OMV vaccine. Furthermore, there was no consistent effect between Opa+ and Opa-OMVs or bacteria, suggesting that the effects of other proteins within these preparations was immunodominant over any Opa effect. This is further supported by a recent study showing no difference in SBA titres following immunisation of CEACAM1-transgenic mice with Opa+ and Opa-OMVs [36]. However, our study and the mouse immunisation study used strain H44/76 and it may be that different Opa proteins have different effects, highlighted by the differences observed between liposomal and recombinant OpaA and OpaD.
One major advantage of this study was the inclusion of a large number of healthy adult subjects. Most previous studies have not stated sample size and the use of buffy coats from the National Blood Service in one study [18] would make it impossible to have specific inclusion and exclusion criteria. Comparison of Opa+ and Opa-strains to determine Opa-dependent    effects is only valid when comparing otherwise isogenic strains, as used in this study. Although most previous studies have also used isogenic strains, one study compared Opa+ OMVs from N. meningitidis with Opa-OMVs from N. lactamica where many other surface proteins would also be different and may exert an immunomodulatory effect. This is the first study of the effects of Opa proteins on CD4 + T cells to include a range of culture conditions, including the use of PBMCs and purified CD4 + T cells, using the same set of antigens with the same measure of proliferation. Different types of Opa+ and Opa-antigens were included to detect a consistent effect of Opa on CD4 + T cells. The lack of inhibitory responses observed with PBMCs ex vivo suggest that Opa proteins would not cause immunosuppressive effects if included as a vaccine component. Opa proteins inhibited CD4 + T cell proliferation after stimulation with IL-2, anti-CD3 and anti-CD28, possibly due to upregulation of CEACAM1 on the T cell surface and/or greater baseline stimulation, which would tend to highlight any inhibitory effects. Future studies would be required to dissect which CD4 + subpopulation is responsible for the effects observed. Opa proteins alone inhibited T cell responses in pre-stimulated in vitro conditions-an effect which was reduced by mutation of the CEA-CAM1-binding regions. However, when included with other antigens in the form of OMVs or killed bacteria, no Opa-dependent inhibition was observed, providing reassurance that current or future meningococcal vaccines containing OMVs would not be negatively affected by Opa protein expression.