Disruption of Aldehyde Reductase Increases Group Size in Dictyostelium *

Developing Dictyostelium cells form structures containing (cid:1) 20,000 cells. The size regulation mechanism involves a secreted counting factor (CF) repressing cytosolic glucose levels. Glucose or a glucose metabolite affects cell-cell adhesion and motility; these in turn affect whether a group stays together, loses cells, or even breaks up. NADPH-coupled aldehyde reductase reduces a wide variety of aldehydes to the corresponding alcohols, including converting glucose to sorbitol. The levels of this enzyme previously appeared to be regulated by CF. We find that disrupting alrA , the gene encoding aldehyde reductase, results in the loss of alrA mRNA and AlrA protein and a decrease in the ability of cell lysates to reduce both glyceraldehyde and glucose in an NADPH-coupled reaction. Counterintuitively, alrA (cid:1) cells grow normally and have decreased glucose levels compared with parental cells. The alrA (cid:1) cells form long unbroken streams and huge groups. Expression of AlrA in alrA (cid:1) cells causes cells to form normal fruiting bodies, indicating that AlrA affects group size. alrA (cid:1) cells have normal adhesion but a reduced motility, and computer simulations suggest that this could indeed result in the formation of large groups. alrA (cid:1)

A fascinating but poorly understood area of biology is how cells create tissues of a specific size (1)(2)(3)(4). A simple model system for this phenomenon is the formation of fruiting bodies in the eukaryote Dictyostelium discoideum, where developing cells form groups of ϳ20,000 cells (see Refs. 5-10 for a review). Dictyostelium normally lives as individual cells that eat bacteria on soil surfaces. As the cells overgrow the bacteria, they starve. The cells then differentiate into either stalk or spore cells and cooperatively form fruiting bodies consisting of a thin, 1-2-mm-high stalk supporting a mass of spores, with the goal of allowing spores to be dispersed by the wind to new patches of soil and a new source of bacteria. For optimal spore dispersal, the fruiting body needs to be as large as possible, with an upper limit dictated by the ability of the stalk to support the spore mass without collapsing or the spore mass sliding down the stalk. Dictyostelium thus have evolved a mechanism to maintain an upper limit to the size of the group of cells that will form an individual fruiting body.
Much of the development of Dictyostelium appears to be regulated by secreted factors. When a cell begins starving, it signals to other cells that it is starving by secreting a glycoprotein called conditioned medium factor (11)(12)(13). As the relative density of starving cells increases, the conditioned medium factor concentration concomitantly increases (14). When the conditioned medium factor concentration passes a threshold concentration, indicating to the population that there is a high density of starving cells, the cells begin aggregating using relayed pulses of extracellular cAMP as a chemoattractant (15)(16)(17)(18).
The aggregating cells form streams that flow toward an aggregation center. If there are more than ϳ20,000 cells in an aggregation stream, the streams break into groups (19). To elucidate how cells sense the presence of too many cells in a stream and how the subsequent morphogenetic reformation occurs, we isolated a shotgun antisense transformant called smlAas that formed very small fruiting bodies due to excessive stream breakup (20). This transformant, as well as smlA Ϫ cells where the corresponding gene was disrupted by homologous recombination, appeared to be oversecreting a factor that when added to starving wild-type cells caused them to form small groups (21). Using this as a bioassay, we partially purified the factor and found that it was a 450-kDa complex of polypeptides we named counting factor (CF) 1 (21). Disruption of countin, a gene encoding one of the subunits of CF, resulted in cells that formed large fruiting bodies due to streams staying intact and coalescing into large groups (22). Diffusion calculations based on the observed accumulation rate of CF indicated that in general the concentration of a secreted factor such as CF could be used to indicate to cells the number of cells in a stream or group (22,23).
Computer simulations of a stream of cells indicated that decreasing cell-cell adhesion and/or increasing random cell motility would cause a stream to dissipate and subsequently fragment (24). If the adhesion then increased or the random motility decreased, the simulations predicted that the dissipated cells would coalesce into a series of groups rather than a single stream. The simulations also predicted that if the adhesion and/or motility were regulated by a secreted factor such as CF, the resulting feedback would allow a very precise control of group size.
We found that CF decreases cell-cell adhesion and increases cell motility (24,25). Decreasing adhesion causes the formation of smaller groups, whereas increasing adhesion or decreasing motility causes the formation of larger groups (24 -27). Together, the observations suggested that, as predicted by the computer simulations, CF regulates group size by regulating adhesion and motility so as to cause a large stream or group to dissipate.
The expression of adhesion molecules and cell motility are regulated by the cAMP pulses that mediate chemotaxis (28 -32). CF potentiates the cAMP-stimulated cAMP pulse and represses a cAMP-induced cGMP pulse (33). The signal transduction pathway that allows CF to regulate cAMP signaling, adhesion, and motility is unknown. High levels of glucose cause cells to form large fruiting bodies (34). CF appears to decrease intracellular glucose levels (35). Increasing glucose levels by adding exogenous glucose negates the effect of CF on group size and mimics the effect of decreasing CF on the cAMP-stimulated cAMP pulse, adhesion, and motility. This suggested that CF might regulate glucose levels to regulate group size.
Two-dimensional gels of aggregating cells showed a prominent spot that seemed to be most intense in countin Ϫ cells and least intense in smlA Ϫ cells (25). Two amino acid sequences obtained from tryptic peptides were used to identify sequence fragments from the Dictyostelium sequencing project. When these fragments were assembled, an open reading frame was identified that coded for a predicted protein with strong similarity to aldehyde reductase. The closely related protein aldose reductase converts glucose and NADPH to sorbitol and NADP (36). Both aldose reductase and aldehyde reductase reduce a wide variety of aldehydes, and their exact functions within cells are still unknown. Aldose reductase is thought to be responsible for some of the complications of diabetes such as neuropathy, cataracts, and retinopathy (37,38). The high levels of glucose in diabetics cause the production of high levels of sorbitol. Sorbitol acts as an osmolyte, and the high levels of sorbitol are then thought to cause a high osmotic pressure within lens and retinal cells, which then causes cellular damage. Because glucose seems to play a role in a cell number-counting signal transduction pathway and because aldose reductase and aldehyde reductase may affect glucose levels, we have examined the function of aldehyde reductase in Dictyostelium cells.

MATERIALS AND METHODS
Sequence Assembly-Preliminary sequence data were obtained from the Dictyostelium BLAST site (available on the World Wide Web at dicty.sdsc.edu/) using the raw reads and contigs provided by the Baylor Sequencing Center and the Institute of Biochemistry (Cologne, Germany) together with the Institute of Molecular Biotechnology (Jena, Germany) and the EUDICT consortium. Sequences were assembled and analyzed using software from the Genetics Computer Group (Madison, WI).
Genomic DNA Extraction-2 ϫ 10 7 vegetative cells were collected by centrifugation and resuspended in 0.5 ml of GL buffer (120 mM NaCl, 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, and 0.5% SDS). 5 l of an RNase mixture (50 units/ml RNase A and 100 units/ml RNase T1) (Roche Applied Science) was added to the cells and incubated for 20 min at 55°C. 50 l of 10 mg/ml proteinase K (Roche Applied Science) was added, and the mixture was incubated for an additional 2 h at 55°C. An equal volume of 1 M Tris (pH 8.0)-buffered phenol was added, and the mixture was gently vortexed. After centrifugation at 10,000 ϫ g for 5 min, the aqueous phase was similarly treated with phenol/chloroform/ isoamyl alcohol (25:24:1) and chloroform/isoamyl alcohol (24:1). The aqueous phase was mixed with 0.1 volumes of 3 M sodium acetate (pH 5.2) and 1 volume of isopropyl alcohol. After centrifugation at 13,000 ϫ g for 15 min, the pellet was washed with 70% ethanol, dried, and resuspended in 50 l of TE (10 mM Tris-HCl, pH 8.0, and 1 mM EDTA, pH 8.0).
Disruption of alrA by Homologous Recombination-PCR was done using Ax4 genomic DNA as a template to create DNA fragments flanking the aldehyde reductase gene. The left arm primers were TCGCGC-CGCGGCTCAACTTCACCAGTTGTCATTTTACC, which added a SacII site to the sequence, and TATCGGCCGAAAACGATGTTCTTGTGT-GTTTGG, which added an EagI site. This generated a 1.0-kb band. The right arm primers were ATCGCGTCGACGTGAAGATAATGGTGCA-CAC, adding a HincII site, and ATAGGGCCCGAGCTGGTCTA-AGGGG, adding an ApaI restriction site. This generated a 1.2-kb fragment. The backbone of the construct used to generate the aldehyde reductase knockout was pBB, pBluescript SK(ϩ) (Stratagene, La Jolla, CA) containing a 1.4-kb blasticidin resistance cassette inserted into the XbaI-HindIII sites (39,40). The left arm fragment generated by the above PCR was digested with SacII and EagI, and the right arm fragment was digested with HincII and ApaI. These fragments were then ligated into the corresponding sites of pBB to generate pARKO. The SacII-ApaI fragment of pARKO was then used to transform Ax4 cells, and blasticidin-resistant cells were selected following Kuspa and Loomis (41). The resulting cell strain was designated ARKO.
Expression of AlrA in alrA Ϫ Cells-To express AlrA in Dictyostelium cells, the expression vector pDXA3-C (42) was altered by Dr. William Deery to remove the ATG upstream of the cDNA insertion site. The sequence of the altered vector from the HindIII site (underlined) to the KpnI (italicized) site is 5Ј-AAGCTTAAAGTTCGAATTCAAAGGTACC-3Ј, and this vector was designated pDXA3-D. A SMART RACE cDNA amplification kit (Clontech Laboratories, Palo Alto, CA) was used to generate cDNA from Ax4 RNA. This cDNA was then used to obtain alrA cDNA using the forward primer 5Ј-GGGGTACCATGGAACCATCATT-TAAATTATC-3Ј (containing a KpnI restriction site and the first 23 nucleotides of the coding region of alrA) and the reverse primer 5Ј-CCCTCGAGTTAATTGAAAAGTGGTACACCC-3Ј (containing an XhoI restriction site and the last 19 nucleotides of the coding region and the termination codon of alrA). This cDNA fragment was digested with KpnI and XhoI and ligated into the similarly digested vector. The cDNA in the resultant vector pAlrAOE was then sequenced to verify that there were no errors. alrA Ϫ cells were then transformed with pAlrAOE and the helper plasmid pREP following Manstein et al. (42), and transformants were selected with 2.5 g/ml G418 and then checked to verify blasticidin resistance. The resulting selected strain was designated ARKOA2. Expression of AlrA in the ARKOA2 cells was verified by staining Western blots of whole cell lysates with anti-AlrA antibodies (see below).
Cell Culture, Group Number Assays, and Western Blots-D. discoideum Ax2 and Ax4 wild types, smlA Ϫ (strain HDB7YA) cells, and cells with a disruption of the ctnA gene (strain HDB2B/4) (referred to in this and previous work as countin Ϫ cells) were grown as previously described (21,22). Photography of fruiting bodies followed Brock et al. (43). Streams and aggregates on SM/5 agar plates with Klebsiella aerogenes bacteria were photographed with a Nikon D1 camera with a macro lens. Cells were developed on filter pads as described in Jain et al. (13). Preparation of conditioned starvation medium (CM) and staining of Western blots with anti-CF50 and anti-countin antibodies followed Brock et al. (43). Cell-mixing (synergy) experiments were performed following Brock et al. (21). Recombinant countin was prepared following Gao et al. (44), and recombinant CF50 was prepared following Brock et al. (43). The effect of recombinant proteins on group size was assayed in submerged culture following Brock and Gomer (22).
Reverse Transcriptase-PCR and Northern Blots-RNA was extracted from 2 ϫ 10 7 cells growing in HL5 at a concentration of 2-4 ϫ 10 6 cells/ml using the RNeasy total RNA isolation kit (Qiagen, San Clarita, CA). cDNA was synthesized using the Prostar Ultra HF reverse transcriptase-PCR system (Stratagene, La Jolla, CA) utilizing Maloney murine leukemia virus reverse transcriptase. PCR was done using the cDNA as a template. The primers GCTGTTGAAGTTGCTCTCGAT-GCTG and GTACACCCCAGAATTTAGCTGGATC were used to amplify a 780-bp aldehyde reductase sequence. As a control, primers GATG-GATCACAATAGATATTCAGCAG and TCCGACTGAATGGGGTTT-GCTATCAT were used to amplify a 217-bp sequence from sslA (accession number AAM34288). RNA isolation from vegetative cells and Northern blots was performed following Brock et al. (43). The 780-bp alrA fragment described above was used as a probe.
Anti-aldehyde Reductase Western Blots and Cell Fractionation-A total of 1 ϫ 10 6 cells were collected by centrifugation and resuspended to 50 l in Laemmli sample buffer. Following the ECL protocol (Amersham Biosciences), the blot was then stained with a 1:2,000 dilution of a 1.3 mg/ml solution of affinity-purified antibody made against the aldehyde reductase-specific peptide CWNTFHKKEHVRPALER (Bethyl Laboratories Inc., Montgomery, TX). For cell fractionation, Ax4 cells were collected by centrifugation and either used immediately or starved in shaking culture in PBM (20 mM KH 2 PO 4 /K 2 HPO 4 , 1 mM MgCl 2 , 0.01 mM CaCl 2 , pH 6.1) for 3 h as described above. 1 ϫ 10 9 cells were collected and resuspended to 1 ϫ 10 8 cells/ml in ice-cold MESES buffer (20 mM MES, pH 6.5, 1 mM EDTA, and 0.25 M sucrose). Cells were lysed through a 5.0-m cameo 25N-syringe filter (Osmonics/MSI, Westborough, MA). Cell fractionation was done using centrifugation following Brock et al. (21).
Aldehyde Reductase Assay-A 5.0-m syringe filter was washed with 5 ml of PBM, and then 3 ml of cells at 5 ϫ 10 7 cells/ml were lysed by a single passage of these cells through the filter. For some assays, the lysate was clarified by centrifugation at 19,000 ϫ g for 1 min. The aldehyde reductase activity was determined by the decrease of NADPH absorption at 340 nm (relative to the background absorption at 400 nm) at room temperature using DL-glyceraldehyde, water, or D-glucose as substrates for the enzyme (45). 500 l of 0.135 M Na 2 HPO 4 /KH 2 PO 4 , pH 6.2, 200 l of water, 100 l of 0.73 mM NADPH tetrasodium salt (Sigma), and 100 l of 50 mM mercaptoethanol were added to 100 l of cell lysate. This was placed in a cuvette, and then either 25 l of 0.04 M DL-glyceraldehyde, 4 M D-glucose, or water was added. The enzymatic activity over the course of 2 min was determined for each substrate by Z ϭ (A T ϭ 0 340 Ϫ A T ϭ 0 400) Ϫ (A T ϭ 2 340 Ϫ A T ϭ 2 400). Following Gabbay and Kinoshita (45), 1 unit of activity was defined as the ability to oxidize 1 mol of NADPH/hour. Given an extinction coefficient for NADPH of 6.22 OD/cm/mM, the units of activity were 4,823 ϫ Z. The values for water were then subtracted from those of glyceraldehyde and glucose. Protein concentrations were determined with a Bio-Rad protein assay in comparison with a bovine serum albumin standard dilution curve.
Adhesion and Motility Assays-Ax4 and alrA Ϫ cells were grown and assayed for adhesion following the protocol of Desbarats et al. (46) as modified by Roisin-Bouffay et al. (24), allowing the disaggregated cells 2 min to aggregate before scoring for single cells. To measure motility, midlog phase cells growing in HL5 were collected by centrifugation, resuspended and washed in PBM, and resuspended to 1 ϫ 10 7 cells/ml. 200 l of cells were starved on a filter pad. At 6, 8, 10, and 12 h, cells were harvested and diluted to 2 ϫ 10 5 cells/ml in PBM, and 200 l of cells was placed in a well of an 8-well slide. For 0-h motilities, midlog cells in HL5 were diluted to 2 ϫ 10 5 cells/ml with HL5, and 200 l of cells was placed in a well of an 8-well slide. For all of the time points, cells were allowed to settle for 15 min, and motility was then measured by videotaping the cells following Yuen et al. (16). The approximate distance moved by a cell was measured in 1-min increments over 10 min.
Computer Simulations-The JAVA computer simulations used the program described in Roisin-Bouffay et al. (24) with the following modifications. The aggregation stream at the beginning of the simulation was 2000 cells in length and 17-22 cells in width. For the distribution of cell motilities, we used the actual distribution of motilities observed with either wild-type or alrA Ϫ cells. The cell-cell adhesions were set as the cell-cell adhesions observed with wild-type or alrA Ϫ cells.
Glucose and Osmolality Assays-Glucose levels were measured as described by Jang et al. (35). To measure osmolality, ϳ2 ϫ 10 8 vegeta-  tive cells or cells starved at 5 ϫ 10 6 cells/ml for 6 h in PBM in shaking culture were collected by centrifugation. The pellets were briefly recentrifuged, the remaining supernatant was removed, and the pellets were frozen at Ϫ80°C. The pellets were thawed, and the osmolality of 50 l of a 1:1 mixture of the lysed cells and distilled water was measured with a model 5004 Micro Osmometer (Precision Systems, Natick, MA). A standard curve was constructed using distilled water and dilutions of a 100-mosmol/kg H 2 O standard solution (Precision Systems), and the osmolality of the original cell lysate was calculated. Protein concentrations were measured using a Bio-Rad protein assay and bovine serum albumin for a calibration curve.
Metabolite Analysis-For vegetative cell samples, cells were grown to 1-2 ϫ 10 6 cells/ml in HL5, and for development cells were starved at 5 ϫ 10 6 cells/ml in PBM in shaking culture. A total of 1 ϫ 10 9 cells were collected by centrifugation at 1,500 ϫ g for 5 min. The pellets were resuspended in ϳ5 ml of the remaining supernatant, and this was recentrifuged. The supernatants were carefully aspirated, and the pellets were frozen at Ϫ80°C. After 1-4 days, the pellets were thawed, and 2 ml of water was added. This mixture was vortexed and then clarified by centrifugation at 23,000 ϫ g for 10 min at 4°C. 1.2 ml of the slightly cloudy supernatant was then frozen on dry ice. After thawing, acetone was added to 50%, and the samples were processed for azetropic dehydration and trimethylsilylation of compounds as described in Shoemaker and Elliot (47) starting at the acetone addition step. Gas chromatography/mass spectrometry of the processed samples was performed as described by Shoemaker and Elliot (47).

RESULTS
Disrupting Aldehyde Reductase Results in Huge Groups-On a two-dimensional gel of whole cells, there was a major protein whose staining intensity seemed to vary as a function of the extracellular CF concentration (25). After digesting the protein from the spot with trypsin, we obtained the amino acid sequence of two of the tryptic peptides. A search of the Dictyostelium genomic DNA sequence identified an open reading frame encoding both of the peptides (see Ref. 25 and Fig. 1). An NCBI protein-protein BLAST search (available on the World Wide Web at www.ncbi.nlm.nih.gov/blast/) using the deduced amino acid sequence of the open reading frame found a 45% identity to porcine alcohol dehydrogenase [NADP ϩ ] (aldehyde reductase) (48,49) and a 44% identity to human (50) and mouse (51) aldehyde reductase and to a putative Arabidopsis mannose-6-phosphate reductase (NADPH-dependent) protein (gi: 15226502). The NCBI conserved domain search performed in conjunction with the BLAST search found the Dictyostelium protein identified above to be 99.6% aligned with the aldo/keto reductase family and 95.7% aligned with the ARA1 aldo/keto reductase-diketogulonate reductase family signatures. The aldo/keto reductase superfamily is made up of a number of related NADPH-dependent oxidoreductases with wide substrate specificities for carbonyl compounds (50). One of the major members of this family is aldehyde reductase, which uses NADPH as a cofactor to reduce a wide variety of aldehydes to the corresponding alcohols (52). The predicted protein identified above, designated AlrA for aldehyde reductase A, is 33.6 kDa with a pI of 5.95. The spot we observed on two-dimensional gels had an apparent molecular mass of ϳ24 kDa and a pI of ϳ6.0. Thus, although the predicted molecular mass is higher than what we observed, the predicted pI is close to the observed pI. The lower molecular mass either may be due to proteolytic processing of AlrA or may be an artifact of the SDS-polyacrylamide gel, since disparities between the predicted mass and the apparent mass on an SDS-polyacrylamide gel have been observed for other proteins (53)(54)(55)(56). There are no predicted transmembrane domains and no strongly charged regions. A BLAST search of the preliminary directory of Dictyostelium genes using the data base of all predicted proteins plus GenBank TM Dictyostelium sequences found strong matches of AlrA to four putative gene products (Table I). A search of the same Dictyostelium sequences using the aldo-keto family consensus sequence found matches to the same gene products (Table I). Together, the data suggest that there are at least five putative protein sequences in the Dictyostelium genome that resemble aldo-keto reductases and that, of these, AlrA has the highest identity with the NCBI consensus sequence for aldo-keto reductases.
To determine whether this putative aldehyde reductase functions to regulate group size, we used homologous recombination to replace a region of the associated gene (starting 295 bp upstream of the A in the ATG encoding the first methionine in the open reading frame and ending 105 bp downstream of the T in the TAA stop codon) with a blasticidin resistance cassette. We identified a transformant designated alrA Ϫ where PCR indicated that the alrA gene had been disrupted, and reverse transcriptase-PCR indicated that the alrA mRNA was absent  (data not shown). A Northern blot of RNA from vegetative cells probed with a fragment of the alrA cDNA indicated that wildtype cells contain a ϳ0.9-kb alrA mRNA, whereas no band was detected in alrA Ϫ cells ( Fig. 2A). A Western blot of total cell lysates indicated that the AlrA protein was detectable in wildtype cells as a 33-kDa band and that this band was absent in the alrA Ϫ cells (Fig. 2B). A standard assay for aldehyde reduc-tase activity is to measure the ability of a cell lysate to oxidize NADPH to NADP in response to an aldehyde such as glyceraldehyde. As shown in Fig. 3A, vegetative (0-h) wild-type cells have a low level of aldehyde reductase activity; this increases at 2 h of starvation and then decreases. Compared with the activity of wild-type cells, the aldehyde reductase activity of alrA Ϫ cells was somewhat less at 0 h and considerably less at 2, 4, and 6 h. The alrA Ϫ cells also had a lower activity than wild-type cells using the aldose glucose as a substrate (Fig. 3B). Together, the data indicate that the alrA Ϫ cells lack the alrA mRNA and the AlrA protein and have reduced NADPH-coupled aldehyde reductase and aldose reductase activities.
Growing alrA Ϫ cells appeared grossly normal by light microscopy and grew as fast as parental cells. Compared with wild-type parental cells (Fig. 4A), developing alrA Ϫ cells formed huge unbroken streams and huge groups (Fig. 4B). Closer examination of the groups showed that whereas wildtype cells formed fruiting bodies (Fig. 4C), the alrA Ϫ cells formed huge mounds of cells (Fig. 4D) that generally did not appear to progress to the formation of fruiting bodies, although occasionally smaller groups of alrA Ϫ cells did form large fruiting bodies.
The large groups formed by the alrA Ϫ cells could be due to the presence of a second mutation or to an effect of the disruption on a nearby gene. To check this, the coding region of the alrA cDNA was expressed under the control of an actin 15 promoter with this construct on a Dictyostelium plasmid. This plasmid was then transformed into alrA Ϫ cells. The resulting alr Ϫ /actin15::alrA cells formed normal sized and even some small fruiting bodies (Fig. 5). This suggests that the absence of AlrA from cells causes them to form large groups.
To determine when AlrA is present during development, a Western blot of vegetative and developing wild-type cells was stained with anti-AlrA antibodies. AlrA is present in vegetative cells; the levels decrease slightly upon starvation and remain roughly constant until 10 h of development and then decline until at 25 h there is very little detectable protein (Fig. 6). Multiple attempts with a wide variety of fixatives failed to yield cells that could be stained with the anti-AlrA antibodies by immunofluorescence (data not shown). As shown in Fig. 7, fractionation of both vegetative and 6-h developing cells by differential centrifugation indicated that AlrA was not detectable in the low speed pellet (containing nuclei (57) and cytoskeletons) and the medium speed pellet (containing large organelles such as mitochondria (58, 59) and lysosomes (60 -63). There is also very little AlrA in the high speed pellet fraction (containing microsomes (64,65), small vesicles (66), and ribosomes (67,68)). AlrA was apparently confined to the high speed supernatant fraction, which contains cytosolic proteins (66). Together, the data suggest that AlrA is present as a cytosolic protein in vegetative and early developing cells.
alrA Ϫ Cells Accumulate Low Extracellular Levels of countin and CF50 -We previously consistently observed a spot on twodimensional gels that was strongest in countin Ϫ cells and weakest in smlA Ϫ cells; the addition of recombinant countin to starving cells decreased the intensity of this spot, whereas the addition of anti-countin antibodies to starving cells increased it (25). After digestion of this spot with trypsin, only two tryptic peptides eluted from the gel. This suggested that there could be other tryptic peptides present in the spot, and for that matter there could be other proteins present in the spot. To determine whether the levels of AlrA are different in countin Ϫ and smlA Ϫ cells, a Western blots of total cell extracts was stained with anti-AlrA antibodies. In a series of experiments, we found that the levels of AlrA in smlA Ϫ , wild-type, and countin Ϫ cells were variable; for instance, sometimes smlA Ϫ cells had more and other times smlA Ϫ cells had less AlrA than parental cells. This suggests that there may be an unknown protein with a pI and a molecular mass similar to those of AlrA and that CF represses the levels of this unknown protein.
One possible reason the alrA Ϫ cells form large groups is that they have a reduced accumulation of extracellular countin and CF50. To check this, Western blots of conditioned starvation medium (CM) were stained for countin and CF50. Compared with the CM from parental wild-type cells, the CM from alrA Ϫ cells had less countin and CF50 (Fig. 8).
alrA Ϫ Cells Are Sensitive to CF-To determine whether AlrA is part of the CF signal transduction pathway, we examined the number of groups formed by cells in the presence or absence of smlA Ϫ cells. When wild-type cells are mixed with smlA Ϫ cells, the high levels of CF secreted by the smlA Ϫ cells cause the FIG. 11. alrA ؊ cells have a reduced osmolality. Vegetative cells and cells starved for 6 h were collected by centrifugation and lysed, and osmolality was measured by freezing point depression. A, osmolality of lysed cells; values are means Ϯ S.E. from three independent experiments. A t test indicated that there is a significant difference between parental and alrA Ϫ cells for both vegetative (veg; p Ͻ 0.005) and starving cells (p Ͻ 0.005). WT, wild type. B, the protein concentration of each sample was measured in duplicate, and the ratio of osmolality to average protein concentration for each sample was calculated; values are means Ϯ S.E. of the ratios from three independent experiments. A t test indicated that there is a significant difference between parental and alrA Ϫ cells for both vegetative (p Ͻ 0.01) and starving cells (p Ͻ 0.025). The average protein concentration Ϯ S.E. in units of mg/ml for the lysed pellets is shown at the bottom of the chart. wild-type cells to form a larger number of groups (21,22). We found that mixing alrA Ϫ cells with 15% smlA Ϫ cells caused the group number to increase by a factor similar to that observed when Ax4 cells were mixed with 15% smlA Ϫ cells (Table II). In addition, the smaller groups that formed when the alrA Ϫ cells were mixed with smlA Ϫ cells generally formed fruiting bodies. The addition of 200 ng/ml recombinant countin to cells caused the group number of parental Ax4 cells and of alrA Ϫ cells to increase. However, the alrA Ϫ group number increased by a somewhat lower percentage (Table II). The addition of 6.3 ng/ml recombinant CF50 also caused the group number of Ax4 and alrA Ϫ cells to increase, but again the alrA Ϫ group number increased by a somewhat lower percentage (Table II). As with mixing with smlA Ϫ cells, adding the recombinant proteins to alrA Ϫ cells increased the number of mounds that formed fruiting bodies. Together, the data suggest that alrA Ϫ cells are sensitive to CF, recombinant countin, and recombinant CF50, although the responses to recombinant countin and recombinant CF50 appear to be less than what is observed in parental cells. This in turn suggests that alrA Ϫ cells have a functional although slightly impaired CF signal transduction pathway.
alrA Ϫ Cells Have Lower Motilities-Computer simulations suggested that increasing cell-cell adhesion and/or decreasing cell motility would cause group size to increase (24,25). As previously observed, the adhesion of wild-type cells increases during development (Fig. 9A) (24, 46, 69 -71). Because of apparently seasonal changes in the behavior of cells and because the reaggregation time in the adhesion assay we have optimized is different from that previously used, the values for the percentage of cells in aggregates are different from the values we previously published (24). However, whether looking at vegetative cells or cells starved for 2, 4, or 6 h, the adhesion of alrA Ϫ cells was roughly similar to that of parental cells.
Besides adhesion, motility can affect group size. There was no significant difference in the motility of vegetative alrA Ϫ and parental cells (Fig. 9B). At 6 and 8 h of development (when streams are forming and breaking up), the alrA Ϫ cells had decreased motilities compared with parental wild-type cells, and then at 10 and 12 h (when the streams have broken and mounds are forming) the motility of alrA Ϫ was again comparable with that of parental cells. At 6 h of development, the distribution of alrA Ϫ cell motilities was skewed toward lower values compared with wild-type cells, but in the alrA Ϫ population there were still some cells with motilities greater than 10 m/min (Fig. 9C). A computer simulation of the behavior of cells in a stream, when the cells were modeled as having the observed adhesion and distribution of cell speeds of either wild-type or alrA Ϫ cells at 6 h, predicted that the alrA Ϫ cells would form groups approximately twice the size of the groups formed by wild-type cells (Fig. 10). This suggested that although the actual cell-cell adhesion decreased slightly in alrA Ϫ cells compared with parental cells, the decrease in motility of alrA Ϫ cells compared with parental cells is enough to increase the group size. Together, the data suggest that disruption of alrA does not have a significant effect on cell-cell adhesion but does significantly decrease motility when the cells are forming and breaking streams and that this decreased motility is predicted to increase group size.
Disruption of Aldehyde Reductase Affects Several Metabolic Pathways-Since aldose reductase is thought to produce the osmolyte sorbitol, we examined the osmolality of alrA Ϫ cells. As shown in Fig. 11A, alrA Ϫ cells have a significantly lower osmolality than parental cells both at the vegetative stage and at 6 h of development. When the protein concentrations of the cell lysates were measured and used to normalize the osmolalities, the alrA Ϫ cells still had a significantly lower osmolality (Fig.  11B). There was no significant difference in the protein concentrations of parental and alrA Ϫ cells, although for both cell lines the protein concentrations increased somewhat at 6 h compared with the vegetative cells. Together, the data indicate that disruption of alrA does not significantly affect the protein concentration in cells but does decrease osmolality.
If the main function of aldehyde reductase was to produce sorbitol from glucose, one might expect that disruption of aldehyde reductase would cause glucose levels to increase. To test this hypothesis, we measured the level of glucose in parental wild-type and alrA Ϫ cells. As shown in Fig. 12, the level of glucose in the vegetative (0 h of starvation) and 6-h starved wild-type cells was similar to what we previously observed for Ax4 cells (35) and somewhat lower than what we previously observed for 2-h starved cells. Compared with the parental cells, the alrA Ϫ cells had lower glucose levels, with a statistically significant difference at 4 h. These results indicate that disruption of alrA causes cells to contain lower rather than higher levels of glucose.
To determine whether the levels of other compounds are affected by disrupting alrA, cells were ruptured by freezethawing, and the soluble components were isolated by centrifugation. After trimethylsilyl derivitization, gas chromatography/mass spectrometry was performed. For vegetative cells, the gas chromatography spectra of the parental and alrA Ϫ extracts were roughly similar (Fig. 13, A and B), although the sizes of some peaks were visibly changed (Fig. 13, A and B, and Table III). For the vegetative cells, the differences were observed as increases in the levels of compounds in the alrA Ϫ extracts. At 6 h of development, the differences between wildtype and alrA Ϫ were mostly observed as increases in the levels of compounds in the alrA Ϫ cells, although alrA Ϫ cells had decreases in the levels of glucopyranose, lactose, and trehalose (Fig. 13, C and D, and Table III). Together, the data suggest that disruption of alrA affect the levels of several metabolites in Dictyostelium.

DISCUSSION
The aldo-keto reductases form a large superfamily of enzymes that reduce CϭO to C-OH in a broad range of aldehydes and ketones (50,72). There are 6 members of the family in bacteria, 14 in yeast (73), and possibly 5 in Dictyostelium. Disruption of alrA in Dictyostelium abolishes the majority of the NADPH-coupled reduction of glyceraldehyde and glucose, suggesting that AlrA is responsible for a large fraction of the aldehyde reductase activity in Dictyostelium cells, at least with respect to these substrates.
Gas chromatography/mass spectroscopy indicated that, compared with parental vegetative cells, vegetative alrA Ϫ cells had increased levels of six identified compounds, all of which contain a carboxylic acid. Given that many organic acids are converted to the corresponding aldehydes and that aldehyde reductase then converts these to the corresponding alcohols, a possible explanation for the increased levels of the organic acids in alrA Ϫ cells is that the conversion to the alcohol is at least partially disrupted and that the levels of the aldehydes increase. This could in turn cause an increase in the levels of the organic acids simply by affecting the equilibrium of an enzymatic reaction or by nonenzymatic oxidation of the aldehydes to the corresponding acids. We did not observe decreased levels of alcohols or any other compound in the gas chromatography/mass spectroscopy analysis of vegetative alrA Ϫ cells. This could be due to lack of reactivity of the alcohols with the derivatizing agent or to a rapid degradation of the alcohols and thus very low levels even in the wild-type cells. In the 6-h starved cells, there were also higher levels of some organic acids in the alrA Ϫ cells. There were also higher levels of a pentose, which could be due to the lack of reduction of a CϭO in the pentose, as well as higher levels of adenosine. Adenosine is not a ketone or an aldehyde, so it is unclear why there are higher levels of this compound in alrA Ϫ cells. The 6-h starved alrA Ϫ cells had lower levels of three different sugars, which could be due to their precursors being aldehydes or ketones.
The physiological function of aldo-keto reductases is unclear, and disruption studies suggest that some individual aldose reductases are not necessary for viability. In yeast, disruption of genes encoding specific aldose reductases tends to affect some but not all of the aldose reductase activity, suggesting that there are overlapping functions and specificities (73). Disruption of the aldose reductase gene in mice has no effect on the viability, gross appearance, or organ morphology (74,75). These mice had slightly elevated serum levels of Ca 2ϩ and Mg 2ϩ , increased urine volume and urinary Ca 2ϩ , and decreased urine osmolality. Disruption of alrA in Dictyostelium had no effect on the growth rate of cells. This is very surprising, since alrA Ϫ cells lack a majority of aldehyde reductase activity and have altered levels of several metabolites. The viability of alrA Ϫ cells could be due to redundancy in the metabolic pathways affected by AlrA, as evidenced by the residual amount of Cells were broken open by freeze-thawing. After removing particulate material by centrifugation, proteins were precipitated with acetone and the remaining material was derivatized with trimethylsilyl. Tracings from gas chromatography of the material are shown and are a representative sample from three independent assays. The material in the numbered peaks was identified by mass spectrometry. The compounds identified by mass spectrometry, referring to the peak numbers (these can vary slightly between gas chromatography runs due to slight temperature differences), are as follows: lactic acid (60) aldehyde reductase observed in the alrA Ϫ cells and by the finding that of the metabolites we could identify by gas chromatography/mass spectrometry, none were completely missing in alrA Ϫ cells.
Disrupting alrA has a profound effect on development, causing the formation of large groups. Expression of the alrA cDNA in the alrA Ϫ cells rescued the group size phenotype, showing that the effect on group size of disrupting alrA was not due to an effect on a nearby gene or a second mutation. However, disruption of alrA does not have an obvious effect on cell growth. One possible explanation for this is that although there is redundancy in the Dictyostelium aldehyde/aldose reductases, the levels of some metabolite affected by the loss of AlrA may not be of significance to growth, as long as there is some of the metabolite present, but this altered level does affect development. Another possible explanation is that AlrA is required for the production or degradation of a compound that is required for or inhibits normal development but that is not required for or does not affect growth.
alrA Ϫ cells form smaller groups when mixed with smlA Ϫ cells and when starved in the presence of recombinant countin or recombinant CF50. This suggests that alrA Ϫ cells have some response to CF and that AlrA is not a key component of the CF signal transduction pathway. Reducing the extracellular levels of either countin or CF50 causes the formation of large groups (22,43), and we found that alrA Ϫ cells have a reduced extracellular accumulation of both proteins. This suggests that one reason alrA Ϫ cells form large groups is that they have a defect in the synthesis, secretion, or processing of extracellular CF. However, the response of alrA Ϫ cells to recombinant countin and recombinant CF50 in terms of reducing group size and increasing group number is not as great as the response of wild-type cells to these two proteins. This suggests that AlrA is required for a complete response of cells to the CF components and that another reason alrA Ϫ cells form large groups is that they have a somewhat attenuated response to CF. Cells lacking either countin or CF50 have high cell-cell adhesion and reduced motilities (22,43). The alrA Ϫ cells have reduced motilities at 6 and 8 h of development, as would be expected from the observation that they have low levels of extracellular countin and CF50. However, the alrA Ϫ cells have roughly normal cell-cell adhesion. This suggests that some effect of disrupting alrA counteracts the effect of low levels of extracellular countin and CF50.
The alrA Ϫ cells have a significantly reduced osmolality compared with parental cells, suggesting that, as in other systems, aldose reductase is involved in regulating Dictyostelium cellular osmolality. Dictyostelium cells respond to an increase in extracellular osmotic strength by increasing levels of cGMP, which causes myosin to be phosphorylated and to redistribute itself to the cell cortex, presumably to strengthen the cell membrane (76). Dictyostelium cells thus appear to use several different mechanisms to regulate and respond to osmotic pressure.
We previously found that CF represses cytosolic glucose levels and that increasing cytosolic glucose levels increases group size (35). This suggested that glucose or some metabolite of glucose might affect group size. The observation that aldose reductase cells form huge groups but do not have abnormally high glucose levels suggests that a metabolite of glucose rather than glucose per se might be a key element of the CF signal transduction pathway.
In rat vascular smooth muscle cells, aldose reductase inhibitors or aldose reductase antisense oligonucleotides block the activation of protein kinase C and the activation of the transcription factor NF-B by some but not all signals, suggesting that aldose reductase activity is necessary for the function of some but not all signal transduction pathways (77). The observation that disruption of alrA in Dictyostelium does not appear to affect growth or adhesion but does affect CF secretion, motility, and group size during development suggests that much remains to be understood about this enigmatic enzyme.

TABLE III
Compounds identified by gas chromatography/mass spectroscopy whose levels are changed in alrA Ϫ cells compared with wild-type cells (Fig. 13) The presence of a compound in two different peaks (for instance glucopyranose in peaks 1239 and 1339) is due to the creation of different trimethylsilyl derivatives.