Definitely do not use AB's MMLV reverse transcriptase - use SuperScript III or QuantaScript instead.  AB's MMLV is one of the weakest RT enzymes out there - giving yields of some transcripts 1000 to 100,000X less than RT enzymes like SSIII and Quanta-script. This, I am fairly confident, is your problem. SSIII and IV are a special mutated more highly thermostable form of MMLV that works much better.

I have already noticed one  noteworthy fact – the more cDNA I use for amplification of these 'problematic' genes – the larger amount of expected product, as compared to occurring non-specific products and primer clouds.

However, it seems that stock cDNA solution (undiluted) is insufficient.

Should I go for larger amount of total RNA in reverse transcription in order to produce adequate amount of cDNA?

Definitely do not use AB's MMLV reverse transcriptase - use SuperScript III or QuantaScript instead.  AB's MMLV is one of the weakest RT enzymes out there - giving yields of some transcripts 1000 to 100,000X less than RT enzymes like SSIII and Quanta-script. This, I am fairly confident, is your problem. SSIII and IV are a special mutated more highly thermostable form of MMLV that works much better.

I would use more RNA for the RT step and as Jack suggested, I would used SuperScript III.  I normally use 1 ug, but with that enzyme you can go higher and still be in the linear range.

Thanks a lot for suggestions.  I am struggling with different amounts of cDNA utilized in reaction,  gradient of primers and annealing temperatures, trying to isolate RNA from different growth stages,  but it seems that issue is far more simple than I thought. 

In this case I will try to purchase SuperScript reverse transcriptase,  but the deadline date is my another problem. 

For the moment,  I will try to perform RT (MMLV)  with 1.0 or maybe even 1.5ug of total RNA. 

Yes, check with your enzyme manufacturer the amount of RNA that particular enzyme (MMLV) can handle properly.

Also, I would advice against using a lot of cDNA in the qPCR reaction.

Try the following: "only 50 ng of RNA was used per reaction, .....and all four nucleotides were at 62.5 µM in the reaction. The reactions were primed by adding

either random hexamers (10 µg) or gene-specific primers to

a final concentration of 50 nM. Aliquots (1 µl) of the reverse

transcription reaction were used in the PCR step. To control for

DNA contamination in the RNA samples, Superscript II RT was

omitted from the RT step. For PCR in this particular experiment,

and throughout the entire work, we used the REDTaq Ready Mix

PCR reaction mix (Sigma) and 0.2 µM of primers. Conditions

for PCR were: 30 cycles of 30 s at 92 ◦C, 30 s at 55 ◦C and 1 min

40 s at 72 ◦C, followed by an extension step of 10 min at 72 ◦C." See in the attached paper.

Article Transcript analysis reveals an extended regulon and the impo...

See Fig 2.7 (p. 55) in the attached.  Panel A is MMLV RT cDNA, and panel B is SSIII cDNA.  In addition, The company's own product literature regarding MMLV cites problems with MMLV RT's ability to bind cDNA (even after denatured) during the PCR/qPCR causing "sporadic results."  You will only run into further problems with MMLV RT unfortunately.

I have done another RT with MMLV by adding 900, 1500, 2100ng of total RNA. Supplier of the kit I am working on advised me to (1.) perform RT in temperature lower than 42°C (going down for 37°C), and (2.) increase the amount of RNAse inhibitor.

I am afraid that I have no time for such experiments, and following Jack’s opinion, it could give no improvement.

Anyway I hope that  next week I will be able to perform trial RT with SS-IV. THANKS A LOT!

Good choice!

Maybe you can try using oligo dT for RT instead of random hexamers. For my case, i found oligo dT is work more better than random hexamers. 

Tiing, I am working on prokaryotes which lack stable poly(A) tail in mRNA. I think random hexamers are suitable in this case.

PS do you think I can use about 80-120U of SSIV in single reverse transcription?

I completely agree with Jack. Additionally, as you know there is no intron in bacterial DNA, and thus reverse transcription readily happens with gDNA that interfere with the gene expression . To avoid this, you must pass your total RNA with DNase-I treatment as to remove gDNA before going for transcriptional reaction. Please follow biorad protocol.

Be careful of using too much SSIV per reaction.  The suggested amount of 10 U/uL in an RT reaction may even be too much.  For a 20 uL RT rxn, use no more than 8 U/uL (e.g. 160 Units total in a 20 uL RT rxn). Your upper value of 120 Units per 20 uL RT rxn would be favorable too.  I assume you are running 20 uL reactions?  If so, your 120 U SSIV per RT rxn would give you 6 Units/uL - which should work very well.

UPDATE:

my problem was fixed by two simple and substantial changes in the protocol.

(1.) additional RNA clean-up with use of DNAse and silica columns

(2.) utilization of efficient reverse transcriptase SuperScript IV.