The secreted Dictyostelium protein CfaD is a chalone

Secreted factors called chalones inhibit the proliferation of the secreting cells, forming a negative feedback loop that can regulate the number of the cells secreting the chalone (Gamer et al., 2003; Gomer, 2001). As the number of cells that secrete the chalone increases, the concentration of the chalone and the associated inhibition of proliferation increases. An example of a chalone is the polypeptide myostatin, which is made by, and secreted from, muscle. As the percentage of the body occupied by muscle increases, the serum concentration of myostatin increases (Lee and McPherron, 1999). Myostatin inhibits myoblast proliferation, and this negative feedback maintains the amount of muscle in the body (Thomas et al., 2000).

For many tissues, the identity of the associated chalone is unclear. In the phenomenon of tumor dormancy, tumors appear to secrete factors that inhibit the proliferation of metastatic cells (Cameron et al., 2000; Guba et al., 2001; Luzzi et al., 1998). Despite the potential use of such factors to inhibit the proliferation of metastases, these factors are largely unknown.

An excellent system to study secreted factors such as chalones is the simple eukaryote Dictyostelium discoideum (Kessin, 2001). Dictyostelium cells normally exist as haploid amoebae that eat bacteria on soil and decaying leaves; laboratory strains can also proliferate in a bacteria-free nutrient broth. When the amoebae starve, they cease to divide and begin to secrete an 80 kDa glycoprotein called conditioned medium factor (CMF). When there is a high density of starving cells, as indicated by a high concentration of CMF (Jain et al., 1992; Yuen et al., 1995), the cells aggregate using relayed pulses of extracellular cAMP as a chemoattractant (Aubry and Firtel, 1999). The aggregating cells form streams that break up into groups of ∼20,000 cells (Shaffer, 1957). Each group develops into a fruiting body consisting of a mass of spore cells supported on a ∼1-mm-high column of stalk cells.

A secreted ∼450 kDa protein complex that is called counting factor (CF) modulates adhesion and motility during aggregation to regulate stream break-up, and thus group- and fruiting-body size (Brock and Gomer, 1999; Gao et al., 2004; Jang and Gomer, 2005; Roisin-Bouffay et al., 2000; Tang et al., 2002). We found that AprA, a 60 kDa protein in a partially purified CF preparation, is not a CF component but, rather, is part of a ∼150 kDa complex that inhibits proliferation and, thus, has the properties of a chalone (Brock and Gomer, 2005). Here, we show that another protein, CfaD, is also not a component of CF. Instead, CfaD is part of a ∼150 kDa complex, interacts with AprA and, similar to AprA, has the properties of a Dictyostelium chalone.

Some preparations of partially purified CF contained a 27 kDa protein. The amino acid (aa) sequence of a tryptic peptide of this protein matched part of an open reading frame in the Dictyostelium genome (supplementary material Fig. S1). We named the predicted protein CfaD for CF-associated protein. The predicted molecular mass of CfaD is 58.6 kDa, suggesting that the 27 kDa protein is a breakdown fragment of CfaD. The predicted CfaD aa sequence contains a peptidase C1A motif and is, over a stretch of 315 aas, 34% similar to cathepsin L precursors from the mosquito Aedes aegypti and other species (supplementary material Fig. S2). Cathepsins are a family of proteases responsible for protein turnover in the lysosome (Nomura and Katunuma, 2005). Tumors often contain increased levels of cathepsins and, unlike normal cells, secrete cathepsins, which appear to promote invasion by degrading the surrounding extracellular matrix (Gocheva and Joyce, 2007; Jedeszko and Sloane, 2004).

CfaD also shows 34% similarity to the 26/29 kDa proteinase of the flesh fly Sarcophaga peregrine (supplementary material Fig. S2), which is synthesized as a ∼62 kDa polypeptide with a 19-aa signal sequence. This protein can hydrolyse the cathepsin substrate Z-Phe-Arg-AMC (Fujimoto et al., 1999). During the processing of the 26/29 kDa proteinase, the signal sequence is removed and the remaining protein is cleaved into a 23 kDa and an ∼25 kDa fragment, whereby the 13 kDa fragment of the precursor that lies between the 23 kDa and 25 kDa fragments is then discarded (Fujimoto et al., 1999). Both the 23 and 25 kDa subunits are post-translationally glycosylated, and the resulting 26 kDa and 29 kDa fragments are secreted by hemocytes into the hemolymph of larvae to degrade the larval midgut and fat body during metamorphosis (Fujimoto et al., 1999; Nakajima et al., 1997; Takahashi et al., 1993). In CfaD, there is a predicted 18-aa signal sequence, and the aa sequence of a tryptic peptide of the secreted form of CfaD begins at the predicted signal sequence cleavage site (supplementary material Fig. S1, arrow), suggesting that the secreted form of the 27 kDa fragment of CfaD (CfaD-27) begins with VPQL.

A comparison of the predicted CfaD aa sequence with other cathepsin sequences (Berti and Storer, 1995; Santamaria et al., 1998) indicated that CfaD contains two key active site residues, a glutamine at position 327 and a cysteine at position 333 (supplementary material Figs S1, S2). However, CfaD belongs to the peptidase C1 family, which includes proteins without peptidase activity (Rawlings and Barrett, 1993). Using the protease assay that showed that the Sarcophaga 26/29-kDa proteinase has a protease activity (Fujimoto et al., 1999), we observed that in PBM (roughly mimicking the extracellular environment), a human cathepsin-L control had activities of ∼2.5 and ∼3.1 nM Z-Phe-Arg-AMC hydrolyzed/hour/μg protein at 22°C and 37°C, respectively. We observed that rCfaD (recombinant CfaD containing a His tag), rHMCfaD (recombinant CfaD containing both His and Myc tags), and rHMCfaD-PM (rHMCfaD with Gln327 changed to Lys and Cys333 changed to Gly) had no detectable protease activity at either 22°C or 37°C, with a detection limit of 0.004 nM Z-Phe-Arg-AMC hydrolyzed/hour/μg protein. In addition, rCfaD had no detectable protease activity at pH 5.2, roughly corresponding to the pH within a lysosome (Aubry et al., 1993). As described below, all three of the rCfaD variants inhibit Dictyostelium cell proliferation, suggesting that CfaD acts as a signal despite its lack of detectable enzymatic activity. This is similar to what we had previously observed for two functional components of the Dictyostelium CF group size regulation signal, CF45-1 (which has similarity to lysozymes but no detectable lysozyme activity) and CF60 (which has similarity to acid phosphatases but little or no acid phosphatase activity) (Brock et al., 2003b; Brock et al., 2006).

To elucidate the function of CfaD, we disrupted cfaD by homologous recombination. A northern blot indicated that wild-type cells contain a 1.9 kb cfaD mRNA, and that there is no detectable cfaD mRNA in cfaD^– cells (Fig. 1A). Affinity-purified anti-CfaD antibodies stained a 65 kDa and a 27 kDa band on western blots of total protein from mid-log phase wild-type cells, and these proteins were not detected in cfaD^– cells (Fig. 1B), which suggests that both proteins are encoded by cfaD, and that the affinity-purified anti-CfaD antibodies are specific for CfaD. Compared with wild type, there were higher levels of both bands in the cells of the CfaD-overexpressing strain cfaD^OE, similar levels in cfaD^– cells that overexpressed CfaD (cfaD^–/cfaD^OE), and somewhat lower levels in cells of the aprA^– strain (that do not express AprA) (Fig. 1B). In other experiments, the levels of CfaD and CfaD-27 (the 27 kDa band) were essentially the same in wild-type and aprA^– cells (data not shown). We had observed previously that the levels of the CF component countin are variable when other CF components are missing (Brock et al., 2003b). Staining cells with affinity-purified anti-CfaD antibodies showed that all vegetative wild-type cells contain CfaD, whereas cfaD^– cells do not show appreciable staining (supplementary material Fig. S3). Deconvolution microscopy indicated that CfaD is concentrated in subcellular structures, possibly vesicles (supplementary material Fig. S3).

Compared with parental wild-type cells, the cfaD^– cells formed large fruiting bodies with large spore heads, whereas cells overexpressing CfaD formed tall fruiting bodies (Fig. 1C). Expression of CfaD in the cfaD^– cells caused these cells to form fruiting bodies that, compared with the cfaD^– fruiting bodies, resembled wild-type fruiting bodies (Fig. 1C). Together, the data suggest that lack of, or overexpression of, CfaD affects development, and that to a first approximation, expression of CfaD in the cfaD^– cells rescues the phenotype, suggesting that the phenotype of the cfaD^– cells is due to disruption of cfaD.

Proliferation curves for cells growing in liquid shaking culture indicated that cfaD^– cells proliferate faster than wild-type cells and reach stationary phase at a significantly higher cell density, whereas cfaD^OE cells proliferate slower and reach stationary phase at a lower density (Fig. 2A,B). The cfaD^–/cfaD^OE cells, which have roughly as much CfaD as wild-type cells, showed proliferation rates and saturation densities that were roughly comparable with those of wild-type cells. From day 1 to day 3, the average doubling times were 12.4 hours for wild-type cells, 10.2 hours for cfaD^– cells, 17.7 hours for cfaD^OE cells, and 11.0 hours for cfaD^–/cfaD^OE cells (Fig. 2B). After reaching saturation density, cfaD^– cells died off faster than wild type or cfaD^–/cfaD^OE cells, whereas cfaD^OE cells were still alive when wild-type or cfaD^–/cfaD^OE cells had died (Fig. 2A). When grown on plates spread with bacteria, there were no significant differences in the proliferation of wild-type cells and the three transformants (Fig. 2C). By contrast, aprA^– cells proliferate faster than wild-type cells on bacteria, suggesting that CfaD and AprA have different functions. The observed doubling times for wild-type cells in shaking culture and on bacteria were similar to those observed previously (Brock and Gomer, 2005; Loomis, 1982).

The evolutionary advantage for Dictyostelium to have AprA appears to be that, although it slows proliferation, it increases spore viability (Brock and Gomer, 2005). We observed that cells lacking CfaD also form structures with a reduced spore count and reduced spore viability, and that expressing CfaD in the cfaD^– background partially rescues both defects (Table 1). In this and a previous report using this assay (Brock and Gomer, 2005), we observed that only ∼1/3 of wild-type spores are viable, possibly due to the detergent used to wash the spores in the assay. Nonetheless, our results suggest that, like AprA, CfaD confers an evolutionary advantage to Dictyostelium cells because it increases spore viability.

Cells lacking AprA proliferate faster than wild-type cells, and tend to be multinucleate (Brock and Gomer, 2005). Compared with those of wild type, cfaD^– cells also tended to be multinucleate, whereas cfaD^–/cfaD^OE cells had nuclei numbers similar to those in wild-type cells (Table 2). We did not observe wild-type or cfaD^–/cfaD^OE cells with more than four nuclei, whereas some cfaD^– cells had as many as eight nuclei. This effect was also seen for cells growing in HL5 on a plastic surface (data not shown).

The growth (the increase in mass or protein per hour) and the proliferation (the increase in the number of cells per hour) of cells can be regulated independently (Dolznig et al., 2004; Gomer, 2001; Jorgensen and Tyers, 2004; Saucedo and Edgar, 2002). The absence of CfaD did not appear to affect mass or protein content of cells (Table 2). The values for wild-type cells are in agreement with previously reported values (Ashworth and Watts, 1970). After normalizing to the number of nuclei, on average cfaD^– cells have less mass and protein per nucleus than wild-type or cfaD^–/cfaD^OE cells (Table 2). Since cells will roughly double their mass in one doubling time, a rough estimate of the growth rate can be obtained by dividing the cell mass or protein content by the doubling time. On a per-cell basis, cfaD^– cells accumulate more mass and nuclei per hour than wild-type cells, but do not have a significantly higher protein accumulation (Table 3). When the growth was calculated per nucleus, there was no significant difference in the mass or protein accumulation rate between cfaD^– and wild type (Table 3). Together, the data suggest that, although cells that lack CfaD have a shorter mitotic cycle, proliferate faster, and on a cell basis accumulate more mass per hour than do wild-type cells, the increased growth rate is due to the increased nuclear and cellular proliferation and is not due to an increased mass or protein accumulation per nucleus.

CfaD-27 accumulates in conditioned growth medium from cells at ∼5×10⁵ cells/ml, a relatively low density, and then at densities above ∼5×10⁶ cells/ml CfaD-staining bands at 60, 55, and 37 kDa appear (Fig. 3B insert). All of these bands were absent in conditioned media from cfaD^– cells, suggesting that the 60 kDa band is CfaD and the other bands are CfaD breakdown products (data not shown). Using known quantities of recombinant CfaD (Fig. 3A) as a standard, at a density of 1.2×10⁷ cells/ml, there was ∼84 ng/ml of CfaD, corresponding to ∼7×10^–6 ng CfaD per cell (Fig. 3B). Since the anti-CfaD antibodies are directed against the entire protein, we were unable to quantify the amount of CfaD-27.

Molecular-sieve fractionation of conditioned growth medium and conditioned starvation medium followed by staining western blots of the fractions indicated that CfaD is present as a ∼150 kDa complex and CfaD-27 as a ∼115 kDa complex in both media (Fig. 4). We previously observed that wild-type-conditioned growth medium contains a broad ∼150 kDa peak of activity that inhibits proliferation, and that this activity was not present in the conditioned growth medium from aprA^– cells (Brock and Gomer, 2005). This suggests that the 150 kDa peak is the major peak of proliferation-inhibiting activity in wild-type cells.

The loss of AprA decreases the apparent molecular mass of the CfaD-containing complex by 25 kDa and the CfaD-27-containing complex by 35 kDa (Fig. 4). However, these are less than the 60 kDa molecular mass of AprA. When using a column that had been used previously for several purifications, we had previously reported that in wild-type-conditioned starvation medium AprA eluted as a broad peak at ∼150 kDa (Brock and Gomer, 2005). When using a new column of the same type, we observed a sharper peak at ∼138 kDa (Fig. 4). The loss of CfaD decreases the apparent molecular mass of AprA by 30 kD, which is less than the molecular mass of CfaD, although similar to the molecular mass of CfaD-27. This suggests a physical link between AprA and CfaD. In addition, the apparent size of all of the complexes described above are smaller than that of the ∼450 kDa CF, using the CF component CF60 as a marker (Brock et al., 2006) (and data not shown). This indicates that CfaD and CfaD-27 are not components of CF.

As an alternative way to determine whether there is an interaction between CfaD and AprA, we carried out pull-down assays. As shown in the upper left panel of Fig. 5, when rHMCfaD was added to either precleared wild-type- or aprA^–-conditioned growth medium together with nickel beads, rHMCfaD was present in the pull-down samples (the material that bound to the nickel beads). When western blots of the pull-down samples were stained with anti-AprA (Fig. 5, bottom left panel), AprA was present in the samples from wild-type- but not aprA^–-conditioned growth medium. This suggests that AprA binds to the rHMCfaD. Similarly, we were able to pull down CfaD by using rAprA (Fig, 5, right panels), further suggesting that AprA and CfaD interact with each other. Neither AprA nor CfaD were pulled down by the beads alone.

The wild-type Dictyostelium strain used in these studies is an axenic strain derived from an isolate from North Carolina called NC4 (Sussman and Sussman, 1967). To determine whether NC4 cells also secrete AprA and CfaD, we grew NC4 cells on a lawn of bacteria on an agar plate, washed off the cells and bacteria, and analysed a solubilized part of the agar by western blotting (see Fig. 6A). As shown in Fig. 6A, NC4 cells secrete both AprA and CfaD into the agar. From 10 μl of agar, there was approximately 0.3 ng of CfaD (Fig. 6A). Measuring the diameter of and thickness of the agar in the plate, we can thus estimate that when there are ∼3×10⁷ cells on the plate, the agar contains ∼850 ng CfaD. This then corresponds to an accumulation of 2.8×10^–5 ng/cell, higher than the accumulation per cell for the axenic wild-type strain in shaking culture. When NC4 cells were grown in shaking culture with bacteria, the conditioned growth medium contained both AprA and CfaD, and molecular-sieve chromatography of this material showed a peak of both proteins at ∼138 kDa (Fig. 6B). Together, the data suggest that CfaD and AprA are secreted by cells in the natural environment to slow proliferation.

To test the hypothesis that CfaD acts as an extracellular signal that inhibits proliferation, recombinant CfaD was added to cells in growth medium. The recombinant CfaD inhibited the proliferation of wild-type and cfaD^– cells (Fig. 7 and Table 4). CfaD appears to slow but not completely inhibit the proliferation of wild-type cells, because we observed that 640 ng/ml recombinant CfaD slowed proliferation to 75±2% of control. For unknown reasons, the maximal inhibition for cfaD^– cells appears to be slightly less than that for wild-type cells. Recombinant CfaD had no observable effect on the proliferation of aprA^– cells (Fig. 7), suggesting that AprA is necessary for the ability of extracellular CfaD to inhibit proliferation. CrlA has similarity with G-protein-coupled receptors, and crlA^– cells proliferate faster than wild-type (Raisley et al., 2004). Recombinant CfaD inhibited the proliferation of crlA^– cells, although the maximal inhibition was less than that for wild-type or crlA^– cells, suggesting that CrlA is not necessary for the effect of CfaD on cells, but does potentiate its activity. For unknown reasons, the EC50 for recombinant CfaD to inhibit proliferation was lower in cfaD^– and crlA^– cells than in wild-type cells (Table 4). Fitting the data to a sigmoidal dose-response curve with a variable Hill coefficient gave a Hill coefficient of 1, indicating that there was no cooperativity in the dose-response curve. To determine whether mutating the putative cathepsin active site of CfaD affects the bioactivity of CfaD, we added rHMCfaD-PM to wild-type cells. At a final concentration of 150 ng/ml, rHMCfaD decreased proliferation at 12 hours by 23±4% (mean ± s.e.m., n=4), whereas rHMCfaD-PM decreased proliferation by 25±2%. Together, the data suggest that, CfaD acts as an extracellular signal that reduces cell proliferation, AprA is necessary for this effect, rHMCfaD and rCfaD have similar bioactivities, and the putative cathepsin-active site of CfaD is not necessary for its ability to slow proliferation.

In this report, we found that CfaD is an autocrine secreted factor that slows cell proliferation, and thus has the properties of a chalone. At high levels of exogenous recombinant CfaD, the proliferation of wild-type cells was slowed by 31% over 12 hours. Given a 12.4 hour doubling time, this corresponds to changing the doubling time to ∼25 hours. Similarly, the 23% decrease in the proliferation of cfaD^– cells over 12 hours corresponds to changing the doubling time from 10.2 hours to ∼16 hours. The increased doubling times roughly correspond to the observed 17.7-hour doubling time for cfaD^OE cells.

Yarger et al. had described a secreted factor that inhibits proliferation at stationary phase (Yarger et al., 1974), has a molecular weight of less than 10 kDa and is heat stable. Since CfaD and AprA are large proteins, the AprA-CfaD complex is probably not identical with this factor. In addition, the factor described by Yarger appears at stationary phase and seems to completely stop proliferation, whereas we observed that AprA (Brock and Gomer, 2005) and CfaD only slow proliferation. This suggests that Dictyostelium cells use the AprA-CfaD complex to slow proliferation as the cells approach saturation, and use the factor described by Yarger to completely stop proliferation when cells reach stationary density.

CfaD and AprA also appear to have different properties. First, AprA inhibits the proliferation of cells growing on bacteria (Brock and Gomer, 2005), whereas CfaD does not. One explanation might be that AprA and CfaD interact with different receptors and signal transduction pathways, even though they are in the same complex. We previously observed this happening for countin and CF50, two components of the CF complex (Brock et al., 2003a; Brock et al., 2002). It is unclear why Dictyostelium cells would have a multi-protein chalone. The observation that recombinant CfaD does not slow proliferation when AprA is absent suggests that slowing proliferation requires the presence of both the AprA and the CfaD signal. Since Dictyostelium cells live in dirt, where there are presumably a large number of different compounds that could activate a receptor, one possibility is that having two different proteins function as signals would be the equivalent of having a message authenticator, decreasing the possibility of an exogenous compound `accidentally' triggering a decrease in proliferation at an inappropriate time.

Compared with cells that contain CfaD, cells that lack CfaD proliferate faster but die faster in shaking culture and have reduced spore viability. One possible reason for the reduced viability of cfaD^– cells is that the decreased amount of mass and protein per nucleus compared with wild-type cells represents less nutrients per nucleus; since protein synthesis per nucleus is not affected by the lack of CfaD, the cfaD^– cells will thus run out of nutrients – especially amino acids – sooner than wild-type cells. The reduced spore viability might also be due to a similar reduction in the amount of nutrients per nucleus. Assuming that the key function of CfaD is to slow proliferation of vegetative cells, the advantage for Dictyostelium cells to use a chalone to slow proliferation would be increased fitness of cells when they are at densities where they may begin to starve.

Cell culture was done following Brock et al. (Brock et al., 1999) in HL5 medium (Formedium Ltd, Norwich, England) using wild-type Ax2 cells, aprA^– strain DB60T3-8 (Brock and Gomer, 2005), and crlA^– strain JH557 (Raisley et al., 2004). Conditioned growth and starvation media (CM) were prepared and concentrated, and PBM buffer was made, following Brock et al. (Brock et al., 2002). Size fractionation was carried out as described in (Brock and Gomer, 2005). Western blots of fractions were stained with anti-AprA antibodies as described in (Brock and Gomer, 2005). To examine the size of the CfaD complex secreted by NC4 cells, 1×10⁶ NC4 cells were grown with live Klebsiella bacteria in PBM in a shaking suspension culture. As a control, bacteria were grown without Dictyostelium cells. After 36 hours, the Dictyostelium cell density was ∼3×10⁷ cells/ml, and the supernatant was clarified and used for gel filtration. Photography of aggregates and fruiting bodies was performed as described in Brock et al. (Brock et al., 2002). Proliferation assays, calculation of doubling times, staining of nuclei, spore counts and spore viability assays were done as described in (Brock and Gomer, 2005).

To generate a homologous recombination cfaD-knockout construct, PCR was performed using Ax2 genomic DNA as a template. All DNA fragments were ligated into pCR 2.1 (Invitrogen, San Diego, CA) and sequenced at Lonestar Labs (Houston, TX). PCR with the primers 5′-CGATAATCATCCGCCGGTATTAGGCCAAGCTCAC-3′ and 5′-GCATGCTCTAGACCTGGGGTAGTGGTACAAACC-3′ yielded a 1138 bp fragment of the 5′ side of cfaD. This was digested with SacII and XbaI, and ligated into the same sites in pBluescript SK+ (Stratagene, La Jolla, CA) which had been previously modified to contain the 1.4 kb SmaI Cre-loxP blasticidin resistance cassette from pLPBLP (Faix et al., 2004) to generate pcf27/29-L. PCR was then carried out with 5′-GCAAATGTAAGCTTGTCTCGCCACCGAGTCCAAC-3′ and 5′-CGCATTGGGCCCGGTTGGATATCAATCAAATCATTATC-3′ to generate a 1056 bp fragment of the 3′ end of cfaD. The fragment was digested with HindIII and ApaI and ligated into the same sites in pcf27/29-L to generate pcf27/29-LR. This was digested with SacII and ApaI, and the insert was purified by gel electrophoresis and a Geneclean II kit (Qbiogene, Inc., Carlsbad, CA). Dictyostelium Ax2 cells were transformed with the construct as described by Shaulsky et al. (Shaulsky et al., 1996). PCR and northern blot analysis were used to verify the disruption of cfaD. Seven cfaD disruption clones with the same phenotype were identified, and all of the results show data from clone DB27C-1, which is referred to in this report as cfaD^–. RNA isolation and Northern blots were done following Brock et al. (Brock et al., 2002). The cDNA encoding the full-length secreted CfaD protein was used as a probe.

To obtain a CfaD-overexpressing construct, PCR was carried out using a vegetative cDNA library and the primers 5′-GATACCGAGCTCATGAATAAATTCATTTTATTATTATC-3′ and 5′-CAGCATCTCGAGTATTCTTTGTTGGAATTGG-3′ to generate a fragment of the cfaD-coding region corresponding to the entire polypeptide starting with the first methionine, and a SacI site on one end and an XhoI site on the other to allow expression of a C-terminal Myc tag. After digestion with SacI and XhoI, the PCR product was ligated into the corresponding sites of pDXA-3D (Ehrenman et al., 2004) to produce the overexpression construct. Ax2 cells were transformed following Manstein et al. (Manstein et al., 1995), and expression of CfaD was verified by staining western blots of whole-cell lysates using anti-CfaD antibodies. The resulting CfaD-overexpressing strain was designated cfaD^OE. Constructs were also made to express the 27 kDa and 29 kDa subunits separately, but only the full-length construct was successfully expressed in our hands. CfaD was similarly expressed in cfaD^– cells, and the resulting strain was designated cfaD^–/cfaD^OE.

Recombinant CfaD was prepared following the method used to prepare recombinant CF50 (Brock et al., 2002) with the exception that 5′-CTTATTCATATGGTTCCACAACTCCCAGCTGC-3′ was used as the forward primer and 5′-CGGATCCTCGAGTTAATTCTTTGTTGGAATTGG-3′ was used as the reverse primer for the PCR reaction. This resulted in a cDNA fragment encoding the region from the first aa of the putative secreted CfaD protein to the TAA stop codon. The resulting recombinant protein was designated rCfaD. Bethyl Laboratories (Montgomery, TX) used this protein to produce affinity-purified rabbit polyclonal anti-CfaD antibodies. Staining of western blots was carried out according to Brock et al. (Brock et al., 2002) using the affinity-purified anti-CfaD antibodies at 0.4 μg/ml.

rCfaD with C-terminal His and Myc tags was generated in to facilitate pull-down assays. Following (Brock et al., 2002), the primers 5′-CTCGAGGTTCCACAACTCCCAGC-3′ and 5′-TCTAGAGCATTCTTTGTTGGAATTGGATAGG-3′ were used to generate a cfaD fragment corresponding to the secreted form of CfaD. This fragment was cloned into a TA cloning vector, pC2.1 (Invitrogen, Carlsbad, CA) which was then digested using XhoI and XbaI to cut out the cfaD fragment. The fragment was ligated into the XhoI and XbaI sites in pBAD/gIII(A) (Invitrogen) to construct pBAD-CfaD, with which Top-10 E.coli cells (Invitrogen) were transformed. To express the resultant protein (designated rHMCfaD), cells containing the pBAD-CfaD construct were grown overnight at 37°C in LB medium (Invitrogen). The overnight culture was then diluted with LB medium to an OD₆₀₀ of 0.1 and further grown at 37°C. Once the culture reached an OD₆₀₀ of 0.5, it was induced by adding 20% arabinose to a final concentration of 0.1%. After 5 hours of induction, cells were collected by centrifugation at 12,000 g for 15 minutes and resuspended in PBS (1.8 mM KH₂PO₄, 10.1 mM Na₂HPO_4, 137 mM NaCl, 2.7 mM KCl, pH 7.4) with EDTA-free protease inhibitors (Roche, Indianapolis, IN). The collected cells were disrupted using a cell disruptor (EmulsiFlex-05, Avestin, Canada) and 30% N-lauroylsarcosine sodium salt solution (Sigma Aldrich, St Louis, MO) was added to a final concentration of 5%. rHMCfaD was then purified using nickel-agarose beads (Qiagen, Valencia, CA) following the manufacturer's protocol. Using a similar protocol, rAprA was prepared using the primers 5′-CTCGAGATGGATTATGTCAATGCTCCTGAC-3′ and 5′-GAATTCCAGTTGCAGTTGAACTAGCACT-3′ to generate the expression plasmid pBAD-AprA.

Using pBAD-CfaD as a template, the primers 5′-CCCCAGTCAAAGATAAAGGTATTTGCGGTTCAGGTTGGACTTTTGG-3′ and 5′-CCAAAAGTCCAACCTGAACCGCAAATACCTTTATCTTTGACTGGGG-3′ were used in a PCR reaction to generate a mutated plasmid (pBAD-CfaD-PM) wherein Gln327 and Cys333 were replaced with Lys and Gly, respectively. The PCR reaction and transformation was carried out using a QuickChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). The resulting plasmid pBAD-CfaD-PM was sequenced to confirm the two point mutations. The plasmid was then transformed into the Top-10 E.coli cells (Invitrogen) to express and purify the mutated recombinant rCfaD (designated rHMCfaD-PM) as described above.

An Innozyme cathepsin L activity kit (Calbiochem, La Jolla, CA) was used to determine the enzymatic activity of rCfaD, using either PBM or 50 mM MES (pH 5.2) as assay buffers. Human cathepsin L provided in the kit was used as a positive control and BSA as a negative control.

Wild-type and aprA^– cells were grown, starting at 1×10⁶ cells/ml, to a density of 5×10⁶ to 6×10⁶ cells/ml. After centrifugation at 3000 g for 3 minutes to clarify conditioned growth medium (CGM), 10 ml of CGM was passed through a 0.2 μm filter (Millipore, Bedford, MA) and was then concentrated to 1 ml using Amicon 10,000 MWCO ultra-filters (Millipore Corporation, Billerica, MA). To remove the proteins that non-specifically bind to nickel-agarose beads (Qiagen, Valencia, CA), the concentrated CGM was pre-cleared by mixing with 25 μl of nickel-agarose beads (50% slurry in PBS) at room temperature. After 1 hour, the pre-cleared CGM was collected by centrifugation at 17,500 g for 10 minutes. To 500 μl of pre-cleared wild-type or aprA^– CGM, 1 μl of 500 ng/μl rHMCfaD was added together with 25 μl of nickel-agarose beads (50% slurry in PBS) and mixed end-to-end overnight at 4 °C. The beads were then washed five times with 1 ml of PBS and collected by centrifugation at 3000 g for 1 minute at room temperature. The beads were then collected using Zymo-P1 fast-spin columns (Zymos Research, Ontario, Canada) and material bound to the beads was eluted using 50 μl of 1×SDS sample buffer. The samples were heated at 95°C for 5 minutes and loaded onto 4-15% Tris-HCl gels (Biorad laboratories, Hercules, CA). Western blots of the gels were then stained either with a 1:1000 dilution of rabbit anti-Myc (Bethyl Laboratories, Montogomery, TX) or anti-AprA antibodies. Similarly, rAprA was used to pull down CfaD.

Statistical analysis using GraphPad Prism software (GraphPad Prism software, San Diego, CA). Differences between two groups were assessed using the Mann-Whitney U test, or between multiple groups by ANOVA. Significance was defined as P<0.05.

We thank Lisa Kreppel and Alan Kimmel for the gift of pLPBLP, Jeff Hadwiger and Dale Hereld for the crlA^– cells, Darrell Pilling for helpful suggestions. This research was supported in part by the National Science Foundation under Grant No. PHY05-51164 to the Kavli Institute for Theoretical Physics, grant number C-1555 from the Robert A. Welch Foundation, and NIH GM074990.